<?xml version="1.0" encoding="UTF-8"?>
 <rdf:RDF xmlns="http://purl.org/rss/1.0/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:cc="http://web.resource.org/cc/" xmlns:syn="http://purl.org/rss/1.0/modules/syndication/" xmlns:admin="http://webns.net/mvcb/">
  <channel rdf:about="http://pinboard.in">
    <title>Pinboard (mraginsky)</title>
    <link>https://pinboard.in/u:mraginsky/public/</link>
    <description>recent bookmarks from mraginsky</description>
    <items>
      <rdf:Seq>	<rdf:li rdf:resource="https://www.sciencedirect.com/science/article/pii/000510987790022X"/>
	<rdf:li rdf:resource="https://www.fabricatedknowledge.com/p/mythos-and-engels-pause?fbclid=IwY2xjawRPrTNleHRuA2FlbQIxMQBicmlkETE4eDlndjllUld2cjlBTkFzc3J0YwZhcHBfaWQQMjIyMDM5MTc4ODIwMDg5MgABHsMwf89zs3VKJoXCiZg-1RshLfpVCcwbEgNOrrL2erKG5FX6sPEz6qE5axUQ_aem_RVUflDX1Xm89FsiXcKYlDQ"/>
	<rdf:li rdf:resource="https://jax-ml.github.io/scaling-book/"/>
	<rdf:li rdf:resource="https://thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference/"/>
	<rdf:li rdf:resource="https://blog.onepatchdown.net/2026/02/22/are-compilers-deterministic-nerd-version/"/>
	<rdf:li rdf:resource="https://thenewstack.io/claude-code-source-leak/"/>
	<rdf:li rdf:resource="https://archive.is/kxYVF"/>
	<rdf:li rdf:resource="https://aeon.co/essays/the-sovereign-individual-and-the-paradox-of-the-digital-age"/>
	<rdf:li rdf:resource="https://proceedings.mlr.press/v139/yarotsky21a.html"/>
	<rdf:li rdf:resource="https://link.springer.com/article/10.1007/s10472-009-9148-3"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2407.13744"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2503.14337"/>
	<rdf:li rdf:resource="https://link.springer.com/article/10.1023/A:1008364332419"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2409.12179"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/1011.0014"/>
	<rdf:li rdf:resource="https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00306/43545/Theoretical-Limitations-of-Self-Attention-in"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2311.00208"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2401.14953"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2401.14029"/>
	<rdf:li rdf:resource="https://royalsocietypublishing.org/doi/10.1098/rsta.2022.0041"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2305.14699"/>
	<rdf:li rdf:resource="https://link.springer.com/article/10.1007/BF02458575"/>
	<rdf:li rdf:resource="https://dominoweb.draco.res.ibm.com/bbdb25acdb530b5d852574ff004efbec.html"/>
	<rdf:li rdf:resource="https://dominoweb.draco.res.ibm.com/49eae98dc5a21de0852574ff005001c8.html"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2210.10749"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2206.06473"/>
	<rdf:li rdf:resource="https://link.springer.com/article/10.1007/s10867-010-9195-3"/>
	<rdf:li rdf:resource="https://www.youtube.com/watch?v=USF1H70bRl0"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2204.12786"/>
	<rdf:li rdf:resource="https://www.goodai.com/a-conversation-with-michael-levin/"/>
	<rdf:li rdf:resource="https://link.springer.com/article/10.1007/s11229-020-02950-3"/>
	<rdf:li rdf:resource="https://www.microsoft.com/en-us/research/publication/123-platonism-constructivism-computer-proofs-vs-proofs-hand/"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2002.03145"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/1604.04295"/>
	<rdf:li rdf:resource="https://arxiv.org/abs/2012.05965"/>
	<rdf:li rdf:resource="https://deontologistics.co/2019/11/01/tfe-information-and-energy/"/>
	<rdf:li rdf:resource="http://www.sciencedirect.com/science/article/pii/0890540191900524"/>
	<rdf:li rdf:resource="https://www.microsoft.com/en-us/research/publication/proving-programs-robust/"/>
	<rdf:li rdf:resource="http://pages.saclay.inria.fr/olivier.temam/files/eval/DLCPTW2014.pdf"/>
	<rdf:li rdf:resource="http://projecteuclid.org/euclid.bj/1377612856"/>
	<rdf:li rdf:resource="http://www.oracle.com/us/corporate/features/sparc-m7/index.html"/>
	<rdf:li rdf:resource="https://satana.math.illinois.edu/"/>
	<rdf:li rdf:resource="http://metamarkets.com/2011/machine-learning-in-wonderland/"/>
	<rdf:li rdf:resource="http://www.podc.org/dijkstra/"/>
	<rdf:li rdf:resource="http://www.cogsci.uci.edu/~ddhoff/ompref.html"/>
	<rdf:li rdf:resource="http://gilkalai.wordpress.com/2010/11/09/subexponential-lower-bound-for-randomized-pivot-rules/"/>
	<rdf:li rdf:resource="http://johncarlosbaez.wordpress.com/2010/10/12/algorithmic-thermodynamics/"/>
	<rdf:li rdf:resource="http://arxiv.org/abs/1010.2067"/>
	<rdf:li rdf:resource="http://arxiv.org/abs/1007.5354"/>
	<rdf:li rdf:resource="http://arxiv.org/abs/0901.2735"/>
	<rdf:li rdf:resource="http://www.eecs.berkeley.edu/~christos/evol/compevol.htm"/>
	<rdf:li rdf:resource="http://modelingwithdata.org/arch/00000032.htm"/>
	<rdf:li rdf:resource="http://www.icsi.berkeley.edu/cgi-bin/pubs/browse.pl?groupid=000006"/>
	<rdf:li rdf:resource="http://people.csail.mit.edu/madhu/papers/gjs-full.pdf"/>
	<rdf:li rdf:resource="http://terrytao.wordpress.com/2009/08/05/mosers-entropy-compression-argument/"/>
	<rdf:li rdf:resource="http://hunch.net/?p=727"/>
	<rdf:li rdf:resource="http://www.stanford.edu/~montanar/BOOK/book.html"/>
      </rdf:Seq>
    </items>
  </channel><item rdf:about="https://www.sciencedirect.com/science/article/pii/000510987790022X">
    <title>The future of control - ScienceDirect</title>
    <dc:date>2026-05-05T18:12:23+00:00</dc:date>
    <link>https://www.sciencedirect.com/science/article/pii/000510987790022X</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[The development of control is briefly reviewed. It is suggested that ‘modern’ control has two aspects: a mathematical investigation of basic properties of dynamical systems, and the development of algorithmic methods of synthesis. Reasons are given for believing that the first of these will have more enduring value than the second. Algorithmic methods which try to eliminate the skill of the designer are contrasted with alternative methods which accept his skill and make it more productive. It is finally suggested that the impact of computers upon industry may give the opportunity for a similar development of production methods which accept and enhance the skill of manual workers.]]></description>
<dc:subject>to-read control-theory engineering complex-systems computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:5c08fa3aa2c7/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:control-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:engineering"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complex-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://www.fabricatedknowledge.com/p/mythos-and-engels-pause?fbclid=IwY2xjawRPrTNleHRuA2FlbQIxMQBicmlkETE4eDlndjllUld2cjlBTkFzc3J0YwZhcHBfaWQQMjIyMDM5MTc4ODIwMDg5MgABHsMwf89zs3VKJoXCiZg-1RshLfpVCcwbEgNOrrL2erKG5FX6sPEz6qE5axUQ_aem_RVUflDX1Xm89FsiXcKYlDQ">
    <title>Engels' Pause and the Permanent Underclass</title>
    <dc:date>2026-04-19T01:49:23+00:00</dc:date>
    <link>https://www.fabricatedknowledge.com/p/mythos-and-engels-pause?fbclid=IwY2xjawRPrTNleHRuA2FlbQIxMQBicmlkETE4eDlndjllUld2cjlBTkFzc3J0YwZhcHBfaWQQMjIyMDM5MTc4ODIwMDg5MgABHsMwf89zs3VKJoXCiZg-1RshLfpVCcwbEgNOrrL2erKG5FX6sPEz6qE5axUQ_aem_RVUflDX1Xm89FsiXcKYlDQ</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>blogs to-read ai security computation economics technology</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:4b2e91f0df03/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:blogs"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:security"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:economics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:technology"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://jax-ml.github.io/scaling-book/">
    <title>How To Scale Your Model</title>
    <dc:date>2026-04-18T22:14:08+00:00</dc:date>
    <link>https://jax-ml.github.io/scaling-book/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>books ai transformers deep-learning computation GPUs algorithms</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:acea80f9cdad/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:books"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:transformers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:deep-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:GPUs"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference/">
    <title>Defeating Nondeterminism in LLM Inference - Thinking Machines Lab</title>
    <dc:date>2026-04-18T22:13:14+00:00</dc:date>
    <link>https://thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>to-read ai machine-learning determinism large-language-models computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:24aaffeb64a7/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:machine-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:determinism"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:large-language-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://blog.onepatchdown.net/2026/02/22/are-compilers-deterministic-nerd-version/">
    <title>Are Compilers Deterministic?</title>
    <dc:date>2026-04-18T22:12:46+00:00</dc:date>
    <link>https://blog.onepatchdown.net/2026/02/22/are-compilers-deterministic-nerd-version/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>computation complex-systems compilers systems engineering determinism</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:62d59df0c279/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complex-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:compilers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:engineering"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:determinism"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://thenewstack.io/claude-code-source-leak/">
    <title>Inside Claude Code's leaked source: swarms, daemons, and 44 features Anthropic kept behind flags - The New Stack</title>
    <dc:date>2026-04-18T21:30:01+00:00</dc:date>
    <link>https://thenewstack.io/claude-code-source-leak/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>ai large-language-models computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:3ea32b41f990/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:large-language-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://archive.is/kxYVF">
    <title>The Edge of Mathematics - The Atlantic</title>
    <dc:date>2026-04-18T21:21:15+00:00</dc:date>
    <link>https://archive.is/kxYVF</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>have-read ai mathematics computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:7cf3ae0fda6c/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:have-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:mathematics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://aeon.co/essays/the-sovereign-individual-and-the-paradox-of-the-digital-age">
    <title>The sovereign individual and the paradox of the digital age | Aeon Essays</title>
    <dc:date>2025-08-24T17:44:21+00:00</dc:date>
    <link>https://aeon.co/essays/the-sovereign-individual-and-the-paradox-of-the-digital-age</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>to-read ai algorithms computation economics capital decision-making</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:31de9a14c266/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:economics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:capital"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:decision-making"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://proceedings.mlr.press/v139/yarotsky21a.html">
    <title>Elementary superexpressive activations</title>
    <dc:date>2025-04-23T14:37:20+00:00</dc:date>
    <link>https://proceedings.mlr.press/v139/yarotsky21a.html</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[We call a finite family of activation functions \emph{superexpressive} if any multivariate continuous function can be approximated by a neural network that uses these activations and has a fixed architecture only depending on the number of input variables (i.e., to achieve any accuracy we only need to adjust the weights, without increasing the number of neurons). Previously, it was known that superexpressive activations exist, but their form was quite complex. We give examples of very simple superexpressive families: for example, we prove that the family {𝑠𝑖𝑛,𝑎𝑟𝑐𝑠𝑖𝑛} is superexpressive. We also show that most practical activations (not involving periodic functions) are not superexpressive. ]]></description>
<dc:subject>papers to-read neural-networks computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:06a574041a03/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:neural-networks"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://link.springer.com/article/10.1007/s10472-009-9148-3">
    <title>On the Vapnik-Chervonenkis dimension of computer programs which use transcendental elementary operations | Annals of Mathematics and Artificial Intelligence</title>
    <dc:date>2025-04-23T14:28:31+00:00</dc:date>
    <link>https://link.springer.com/article/10.1007/s10472-009-9148-3</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[We exhibit upper bounds for the Vapnik-Chervonenkis (VC) dimension of a wide family of concept classes that are defined by algorithms using analytic Pfaffian functions. We give upper bounds on the VC dimension of concept classes in which the membership test for whether an input belongs to a concept in the class can be performed either by a computation tree or by a circuit with sign gates containing Pfaffian functions as operators. These new bounds are polynomial both in the height of the tree and in the depth of the circuit. As consequence we obtain polynomial VC dimension not also for classes of concepts whose membership test can be defined by polynomial time algorithms but also for those defined by well-parallelizable sequential exponential time algorithms.]]></description>
<dc:subject>papers to-read computation computational-complexity learning-theory</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:f4c23285309c/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computational-complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:learning-theory"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2407.13744">
    <title>[2407.13744] LLMs as Function Approximators: Terminology, Taxonomy, and Questions for Evaluation</title>
    <dc:date>2025-04-15T18:54:02+00:00</dc:date>
    <link>https://arxiv.org/abs/2407.13744</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Natural Language Processing has moved rather quickly from modelling specific tasks to taking more general pre-trained models and fine-tuning them for specific tasks, to a point where we now have what appear to be inherently generalist models. This paper argues that the resultant loss of clarity on what these models model leads to metaphors like "artificial general intelligences" that are not helpful for evaluating their strengths and weaknesses. The proposal is to see their generality, and their potential value, in their ability to approximate specialist function, based on a natural language specification. This framing brings to the fore questions of the quality of the approximation, but beyond that, also questions of discoverability, stability, and protectability of these functions. As the paper will show, this framing hence brings together in one conceptual framework various aspects of evaluation, both from a practical and a theoretical perspective, as well as questions often relegated to a secondary status (such as "prompt injection" and "jailbreaking"). ]]></description>
<dc:subject>papers to-read large-language-models function-approximation automata generative-models computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:c3ee31a65c73/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:large-language-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:function-approximation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:automata"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:generative-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2503.14337">
    <title>[2503.14337] PENCIL: Long Thoughts with Short Memory</title>
    <dc:date>2025-03-20T15:05:33+00:00</dc:date>
    <link>https://arxiv.org/abs/2503.14337</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[While recent works (e.g. o1, DeepSeek R1) have demonstrated great promise of using long Chain-of-Thought (CoT) to improve reasoning capabilities of language models, scaling it up during test-time is challenging due to inefficient memory usage -- intermediate computations accumulate indefinitely in context even no longer needed for future thoughts. We propose PENCIL, which incorporates a reduction mechanism into the autoregressive generation process, allowing the model to recursively clean up intermediate thoughts based on patterns learned from training. With this reduction mechanism, PENCIL significantly reduces the maximal context length required during generation, and thus can generate longer thoughts with limited memory, solving larger-scale problems given more thinking time. For example, we demonstrate PENCIL achieves 97\% accuracy on the challenging Einstein's puzzle -- a task even large models like GPT-4 struggle with -- using only a small 25M-parameter transformer with 2048 context length. Theoretically, we prove PENCIL can perform universal space-efficient computation by simulating Turing machines with optimal time and space complexity, and thus can solve arbitrary computational tasks that would otherwise be intractable given context window constraints. ]]></description>
<dc:subject>papers to-read computation large-language-models computational-complexity</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:2a029727e26e/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:large-language-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computational-complexity"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://link.springer.com/article/10.1023/A:1008364332419">
    <title>When Physical Systems Realize Functions... | Minds and Machines</title>
    <dc:date>2025-02-15T04:02:05+00:00</dc:date>
    <link>https://link.springer.com/article/10.1023/A:1008364332419</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[After briefly discussing the relevance of the notions ‘computation’ and ‘implementation’ for cognitive science, I summarize some of the problems that have been found in their most common interpretations. In particular, I argue that standard notions of computation together with a ‘state-to-state correspondence view of implementation’ cannot overcome difficulties posed by Putnam's Realization Theorem and that, therefore, a different approach to implementation is required. The notion ‘realization of a function’, developed out of physical theories, is then introduced as a replacement for the notional pair ‘computation-implementation’. After gradual refinement, taking practical constraints into account, this notion gives rise to the notion ‘digital system’ which singles out physical systems that could be actually used, and possibly even built.]]></description>
<dc:subject>papers to-read computation philosophy physics systems</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:034cd3252fda/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:physics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:systems"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2409.12179">
    <title>[2409.12179] Computational Dynamical Systems</title>
    <dc:date>2025-01-20T21:17:07+00:00</dc:date>
    <link>https://arxiv.org/abs/2409.12179</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[We study the computational complexity theory of smooth, finite-dimensional dynamical systems. Building off of previous work, we give definitions for what it means for a smooth dynamical system to simulate a Turing machine. We then show that 'chaotic' dynamical systems (more precisely, Axiom A systems) and 'integrable' dynamical systems (more generally, measure-preserving systems) cannot robustly simulate universal Turing machines, although such machines can be robustly simulated by other kinds of dynamical systems. Subsequently, we show that any Turing machine that can be encoded into a structurally stable one-dimensional dynamical system must have a decidable halting problem, and moreover an explicit time complexity bound in instances where it does halt. More broadly, our work elucidates what it means for one 'machine' to simulate another, and emphasizes the necessity of defining low-complexity 'encoders' and 'decoders' to translate between the dynamics of the simulation and the system being simulated. We highlight how the notion of a computational dynamical system leads to questions at the intersection of computational complexity theory, dynamical systems theory, and real algebraic geometry. ]]></description>
<dc:subject>papers to-read computation dynamical-systems</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:443b5d7ab1e5/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:dynamical-systems"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/1011.0014">
    <title>[1011.0014] Galois Theory of Algorithms</title>
    <dc:date>2024-07-27T22:37:00+00:00</dc:date>
    <link>https://arxiv.org/abs/1011.0014</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Many different programs are the implementation of the same algorithm. The collection of programs can be partitioned into different classes corresponding to the algorithms they implement. This makes the collection of algorithms a quotient of the collection of programs. Similarly, there are many different algorithms that implement the same computable function. The collection of algorithms can be partitioned into different classes corresponding to what computable function they implement. This makes the collection of computable functions into a quotient of the collection of algorithms. Algorithms are intermediate between programs and functions:
Programs $\twoheadrightarrow$ Algorithms $\twoheadrightarrow$ Functions.
\noindent Galois theory investigates the way that a subobject sits inside an object. We investigate how a quotient object sits inside an object. By looking at the Galois group of programs, we study the intermediate types of algorithms possible and the types of structures these algorithms can have. ]]></description>
<dc:subject>papers to-read computation programming-languages algorithms</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:553c510d994a/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:programming-languages"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00306/43545/Theoretical-Limitations-of-Self-Attention-in">
    <title>Theoretical Limitations of Self-Attention in Neural Sequence Models | Transactions of the Association for Computational Linguistics | MIT Press</title>
    <dc:date>2024-03-04T21:22:35+00:00</dc:date>
    <link>https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00306/43545/Theoretical-Limitations-of-Self-Attention-in</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Transformers are emerging as the new workhorse of NLP, showing great success across tasks. Unlike LSTMs, transformers process input sequences entirely through self-attention. Previous work has suggested that the computational capabilities of self-attention to process hierarchical structures are limited. In this work, we mathematically investigate the computational power of self-attention to model formal languages. Across both soft and hard attention, we show strong theoretical limitations of the computational abilities of self-attention, finding that it cannot model periodic finite-state languages, nor hierarchical structure, unless the number of layers or heads increases with input length. These limitations seem surprising given the practical success of self-attention and the prominent role assigned to hierarchical structure in linguistics, suggesting that natural language can be approximated well with models that are too weak for the formal languages typically assumed in theoretical linguistics.]]></description>
<dc:subject>papers to-read transformers formal-languages computation automata</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:b4f6c093e181/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:transformers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:formal-languages"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:automata"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2311.00208">
    <title>[2311.00208] Transformers as Recognizers of Formal Languages: A Survey on Expressivity</title>
    <dc:date>2024-01-31T18:46:12+00:00</dc:date>
    <link>https://arxiv.org/abs/2311.00208</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[As transformers have gained prominence in natural language processing, some researchers have investigated theoretically what problems they can and cannot solve, by treating problems as formal languages. Exploring questions such as this will help to compare transformers with other models, and transformer variants with one another, for various tasks. Work in this subarea has made considerable progress in recent years. Here, we undertake a comprehensive survey of this work, documenting the diverse assumptions that underlie different results and providing a unified framework for harmonizing seemingly contradictory findings. ]]></description>
<dc:subject>papers to-read computation formal-methods transformers</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:a82a9d4078b4/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:formal-methods"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:transformers"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2401.14953">
    <title>[2401.14953] Learning Universal Predictors</title>
    <dc:date>2024-01-29T15:23:18+00:00</dc:date>
    <link>https://arxiv.org/abs/2401.14953</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Meta-learning has emerged as a powerful approach to train neural networks to learn new tasks quickly from limited data. Broad exposure to different tasks leads to versatile representations enabling general problem solving. But, what are the limits of meta-learning? In this work, we explore the potential of amortizing the most powerful universal predictor, namely Solomonoff Induction (SI), into neural networks via leveraging meta-learning to its limits. We use Universal Turing Machines (UTMs) to generate training data used to expose networks to a broad range of patterns. We provide theoretical analysis of the UTM data generation processes and meta-training protocols. We conduct comprehensive experiments with neural architectures (e.g. LSTMs, Transformers) and algorithmic data generators of varying complexity and universality. Our results suggest that UTM data is a valuable resource for meta-learning, and that it can be used to train neural networks capable of learning universal prediction strategies. ]]></description>
<dc:subject>papers to-read computation machine-learning universal-prediction neural-networks</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:469fce73ffe8/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:machine-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:universal-prediction"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:neural-networks"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2401.14029">
    <title>[2401.14029] Towards a Systems Theory of Algorithms</title>
    <dc:date>2024-01-26T03:06:17+00:00</dc:date>
    <link>https://arxiv.org/abs/2401.14029</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Traditionally, numerical algorithms are seen as isolated pieces of code confined to an {\em in silico} existence. However, this perspective is not appropriate for many modern computational approaches in control, learning, or optimization, wherein {\em in vivo} algorithms interact with their environment. Examples of such {\em open} include various real-time optimization-based control strategies, reinforcement learning, decision-making architectures, online optimization, and many more. Further, even {\em closed} algorithms in learning or optimization are increasingly abstracted in block diagrams with interacting dynamic modules and pipelines. In this opinion paper, we state our vision on a to-be-cultivated {\em systems theory of algorithms} and argue in favour of viewing algorithms as open dynamical systems interacting with other algorithms, physical systems, humans, or databases. Remarkably, the manifold tools developed under the umbrella of systems theory also provide valuable insights into this burgeoning paradigm shift and its accompanying challenges in the algorithmic world. We survey various instances where the principles of algorithmic systems theory are being developed and outline pertinent modeling, analysis, and design challenges. ]]></description>
<dc:subject>papers to-read control-theory algorithms computation learning optimization</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:3cf791653c0d/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:control-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:optimization"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://royalsocietypublishing.org/doi/10.1098/rsta.2022.0041">
    <title>Symbols and grounding in large language models | Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences</title>
    <dc:date>2023-07-28T16:59:42+00:00</dc:date>
    <link>https://royalsocietypublishing.org/doi/10.1098/rsta.2022.0041</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Large language models (LLMs) are one of the most impressive achievements of artificial intelligence in recent years. However, their relevance to the study of language more broadly remains unclear. This article considers the potential of LLMs to serve as models of language understanding in humans. While debate on this question typically centres around models’ performance on challenging language understanding tasks, this article argues that the answer depends on models’ underlying competence, and thus that the focus of the debate should be on empirical work which seeks to characterize the representations and processing algorithms that underlie model behaviour. From this perspective, the article offers counterarguments to two commonly cited reasons why LLMs cannot serve as plausible models of language in humans: their lack of symbolic structure and their lack of grounding. For each, a case is made that recent empirical trends undermine the common assumptions about LLMs, and thus that it is premature to draw conclusions about LLMs’ ability (or lack thereof) to offer insights on human language representation and understanding.

This article is part of a discussion meeting issue ‘Cognitive artificial intelligence’.]]></description>
<dc:subject>papers to-read ai language cognitive-science computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:e9117ab25de1/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:language"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cognitive-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2305.14699">
    <title>[2305.14699] Can Transformers Learn to Solve Problems Recursively?</title>
    <dc:date>2023-06-03T02:40:07+00:00</dc:date>
    <link>https://arxiv.org/abs/2305.14699</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Neural networks have in recent years shown promise for helping software engineers write programs and even formally verify them. While semantic information plays a crucial part in these processes, it remains unclear to what degree popular neural architectures like transformers are capable of modeling that information. This paper examines the behavior of neural networks learning algorithms relevant to programs and formal verification proofs through the lens of mechanistic interpretability, focusing in particular on structural recursion. Structural recursion is at the heart of tasks on which symbolic tools currently outperform neural models, like inferring semantic relations between datatypes and emulating program behavior. We evaluate the ability of transformer models to learn to emulate the behavior of structurally recursive functions from input-output examples. Our evaluation includes empirical and conceptual analyses of the limitations and capabilities of transformer models in approximating these functions, as well as reconstructions of the ``shortcut" algorithms the model learns. By reconstructing these algorithms, we are able to correctly predict 91 percent of failure cases for one of the approximated functions. Our work provides a new foundation for understanding the behavior of neural networks that fail to solve the very tasks they are trained for. 

--- Talia's Twitter thread: https://twitter.com/TaliaRinger/status/1661786081249964050?s=20]]></description>
<dc:subject>self-promotion transformers programming-languages state-space-models computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:b49617e52da9/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:self-promotion"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:transformers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:programming-languages"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:state-space-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://link.springer.com/article/10.1007/BF02458575">
    <title>The cellular computer DNA: Program or data | SpringerLink</title>
    <dc:date>2023-05-07T20:07:53+00:00</dc:date>
    <link>https://link.springer.com/article/10.1007/BF02458575</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[The classical metaphor of the genetic program written in the DNA nucleotidic sequences is reconsidered. Recent works on algorithmic complexity and logical properties of computer programs and data are used to question the explanatory value of that metaphor. Structural properties of strings are looked for which would be necessary to apply to DNA sequences if the metaphor is to be taken literally. The notion of sophistication is used to quantify meaningful complexity and to distinguish it from classical computational complexity. In this context, the distinction between program and data becomes relevant and an alternative metaphor of DNA as data to a parallel computing network embedded in the global geometrical and biochemical structure of the cell is discussed. An intermediate picture of an evolving network emerges as the most likely where the output of the cellular computing network can produce, at a different time scale, changes in the structure of the network itself by means of changes in the DNA activity patterns.]]></description>
<dc:subject>papers to-read biology genetics complexity computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:1520e370e1a6/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:biology"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:genetics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://dominoweb.draco.res.ibm.com/bbdb25acdb530b5d852574ff004efbec.html">
    <title>IBM Research | Technical Paper Search | A Junction between Computer Science and Category Theory, I: Basic Concepts and Examples (Part 2)(Search Reports)</title>
    <dc:date>2023-02-18T17:54:59+00:00</dc:date>
    <link>https://dominoweb.draco.res.ibm.com/bbdb25acdb530b5d852574ff004efbec.html</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[This is the second part of the first report in a series devoted to exploring the interface or "junction" between computer science and category theory. Both benefit from this exploration: computer science by a powerful set of tools and a general methodoloqy providing a rigorous and uniform approach to many of its basic concepts, methods, and questions; and category theory by a nontrivial collection of practical applications and illustrations, plus a number of new problems and results. Our present general purposes are to provide a clear, leisurely, and well-illustrated introduction to the basic lanquage of category theory, and to give introductory formulations of some of the computer science topics, including programs, machines, automata, and languages.

This Part covers graphs and diagrams, and introduces the third key categorial concept, natural transformation. An extended example covering correctness and termination of flow-diagram programs illustrates many of the concepts covered so far in the series.]]></description>
<dc:subject>papers to-read categories computation computer-science</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:3f93250df79c/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:categories"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://dominoweb.draco.res.ibm.com/49eae98dc5a21de0852574ff005001c8.html">
    <title>IBM Research | Technical Paper Search | A Junction between Computer Science and Category Theory, I: Basic Concepts and Examples (Part 1)(Search Reports)</title>
    <dc:date>2023-02-18T17:54:17+00:00</dc:date>
    <link>https://dominoweb.draco.res.ibm.com/49eae98dc5a21de0852574ff005001c8.html</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[This is the first part of the first report in a series devoted to exploring the interface or "junction" between computer science and category theory. We expect that both will benefit from the exploration: computer science by a powerful set of tools and a general methodology providing a rigorous and uniform approach to many of its basic concepts, methods, and questions; and category theory by a nontrivial collection of practical applications and illustrations, plus a number of new problems and results. Our present general purposes are to provide a clear, leisurely, and well-illustrated introduction to the basic language of category theory, and to give introductory formulations of some of the computer science topics we treat later in greater depth, including machines, automata, and languagea. Later reports will contain the research results which led us to undertake this series.

Section 0 contains a general introduction, a discussion of the special relevance of category theory to computer science, and intuitive interpretations for the key categerical concepts. Section 1 contains a compendium of the background definitions and notation assumed in subsequent sections, especially oriented toward our computer application and category theoretic viewpoint. Section 2 contains the two most basic definitions, of category and functor. There are many examples and frequent intuitive discussions.]]></description>
<dc:subject>papers to-read categories computation computer-science</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:9e8fc749775e/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:categories"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2210.10749">
    <title>[2210.10749] Transformers Learn Shortcuts to Automata</title>
    <dc:date>2023-02-01T18:02:07+00:00</dc:date>
    <link>https://arxiv.org/abs/2210.10749</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Algorithmic reasoning requires capabilities which are most naturally understood through recurrent models of computation, like the Turing machine. However, Transformer models, while lacking recurrence, are able to perform such reasoning using far fewer layers than the number of reasoning steps. This raises the question: what solutions are these shallow and non-recurrent models finding? We investigate this question in the setting of learning automata, discrete dynamical systems naturally suited to recurrent modeling and expressing algorithmic tasks. Our theoretical results completely characterize shortcut solutions, whereby a shallow Transformer with only $o(T)$ layers can exactly replicate the computation of an automaton on an input sequence of length $T$. By representing automata using the algebraic structure of their underlying transformation semigroups, we obtain $O(\log T)$-depth simulators for all automata and $O(1)$-depth simulators for all automata whose associated groups are solvable. Empirically, we perform synthetic experiments by training Transformers to simulate a wide variety of automata, and show that shortcut solutions can be learned via standard training. We further investigate the brittleness of these solutions and propose potential mitigations. ]]></description>
<dc:subject>papers to-read transformers computation machine-learning complexity</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:6326d3642982/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:transformers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:machine-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2206.06473">
    <title>[2206.06473] A Dilemma for Solomonoff Prediction</title>
    <dc:date>2022-09-18T19:38:20+00:00</dc:date>
    <link>https://arxiv.org/abs/2206.06473</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[The framework of Solomonoff prediction assigns prior probability to hypotheses inversely proportional to their Kolmogorov complexity. There are two well-known problems. First, the Solomonoff prior is relative to a choice of Universal Turing machine. Second, the Solomonoff prior is not computable. However, there are responses to both problems. Different Solomonoff priors converge with more and more data. Further, there are computable approximations to the Solomonoff prior. I argue that there is a tension between these two responses. This is because computable approximations to Solomonoff prediction do not always converge. ]]></description>
<dc:subject>papers to-read computation prediction philosophy</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:57e14ac90280/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:prediction"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://link.springer.com/article/10.1007/s10867-010-9195-3">
    <title>Information processing, computation, and cognition | SpringerLink</title>
    <dc:date>2022-08-03T17:15:40+00:00</dc:date>
    <link>https://link.springer.com/article/10.1007/s10867-010-9195-3</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Computation and information processing are among the most fundamental notions in cognitive science. They are also among the most imprecisely discussed. Many cognitive scientists take it for granted that cognition involves computation, information processing, or both – although others disagree vehemently. Yet different cognitive scientists use ‘computation’ and ‘information processing’ to mean different things, sometimes without realizing that they do. In addition, computation and information processing are surrounded by several myths; first and foremost, that they are the same thing. In this paper, we address this unsatisfactory state of affairs by presenting a general and theory-neutral account of computation and information processing. We also apply our framework by analyzing the relations between computation and information processing on one hand and classicism, connectionism, and computational neuroscience on the other. We defend the relevance to cognitive science of both computation, at least in a generic sense, and information processing, in three important senses of the term. Our account advances several foundational debates in cognitive science by untangling some of their conceptual knots in a theory-neutral way. By leveling the playing field, we pave the way for the future resolution of the debates’ empirical aspects.]]></description>
<dc:subject>papers to-read information-theory cognition computation biology neuroscience</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:580f27a1f05d/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cognition"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:biology"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:neuroscience"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://www.youtube.com/watch?v=USF1H70bRl0">
    <title>Brian Cantwell Smith The philosophy of computation meaning, mechanism, mystery - YouTube</title>
    <dc:date>2022-08-02T16:35:23+00:00</dc:date>
    <link>https://www.youtube.com/watch?v=USF1H70bRl0</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>philosophy cognitive-science ai computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:9ffa8e1f0ccc/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cognitive-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2204.12786">
    <title>[2204.12786] Machines of finite depth: towards a formalization of neural networks</title>
    <dc:date>2022-08-02T16:31:20+00:00</dc:date>
    <link>https://arxiv.org/abs/2204.12786</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[We provide a unifying framework where artificial neural networks and their architectures can be formally described as particular cases of a general mathematical construction--machines of finite depth. Unlike neural networks, machines have a precise definition, from which several properties follow naturally. Machines of finite depth are modular (they can be combined), efficiently computable and differentiable. The backward pass of a machine is again a machine and can be computed without overhead using the same procedure as the forward pass. We prove this statement theoretically and practically, via a unified implementation that generalizes several classical architectures--dense, convolutional, and recurrent neural networks with a rich shortcut structure--and their respective backpropagation rules. ]]></description>
<dc:subject>papers to-read neural-networks computation dynamical-systems systems</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:bf8b501d87fa/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:neural-networks"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:dynamical-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:systems"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://www.goodai.com/a-conversation-with-michael-levin/">
    <title>A Conversation with Michael Levin | GoodAI</title>
    <dc:date>2022-08-02T16:03:13+00:00</dc:date>
    <link>https://www.goodai.com/a-conversation-with-michael-levin/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>to-read blogs biology cybernetics ai complex-systems computation adaptive-systems agency</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:9de62de6e858/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:blogs"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:biology"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cybernetics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:ai"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complex-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:adaptive-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:agency"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://link.springer.com/article/10.1007/s11229-020-02950-3">
    <title>The computational philosophy: simulation as a core philosophical method | SpringerLink</title>
    <dc:date>2022-08-02T16:01:32+00:00</dc:date>
    <link>https://link.springer.com/article/10.1007/s11229-020-02950-3</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Modeling and computer simulations, we claim, should be considered core philosophical methods. More precisely, we will defend two theses. First, philosophers should use simulations for many of the same reasons we currently use thought experiments. In fact, simulations are superior to thought experiments in achieving some philosophical goals. Second, devising and coding computational models instill good philosophical habits of mind. Throughout the paper, we respond to the often implicit objection that computer modeling is “not philosophical.”]]></description>
<dc:subject>papers to-read philosophy-of-science simulation complex-systems computation via:cshalizi</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:461a1ef48606/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy-of-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:simulation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complex-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:via:cshalizi"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://www.microsoft.com/en-us/research/publication/123-platonism-constructivism-computer-proofs-vs-proofs-hand/">
    <title>Platonism, Constructivism, and Computer Proofs vs. Proofs by Hand - Microsoft Research</title>
    <dc:date>2022-07-28T03:12:21+00:00</dc:date>
    <link>https://www.microsoft.com/en-us/research/publication/123-platonism-constructivism-computer-proofs-vs-proofs-hand/</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[In one of Krylov’s fables, a small dog Moska barks at the elephant who pays no attention whatsoever to Moska. This image comes to my mind when I think of constructive mathematics versus “classical” (that is mainstream) mathematics. In this article, we put a few words into the elephant’s mouth. The idea to write such an article came to me in the summer of 1995 when I came across a fascinating 1917 bet between the constructivist Hermann Weyl and George Polya, a classical mathematician. An English translation of the bet (from German) is found in the article.

Our main objection to the historical constructivism is that it has not been sufficiently constructive. The constructivists have been obsessed with computability and have not paid sufficient attention to the feasibility of algorithms. However, the constructivists’ criticism of classical mathematics has a point. Instead of dismissing constructivism offhand, it makes sense to come up with a positive alternative, an antithesis to historical constructivism. We believe that we have found such an alternative. In fact, it is well known and very popular in computer science: the principle of separating concerns.

[Added in July 2006] The additional part on computer proofs vs. proofs by hand was a result of frustration that many computer scientists would not trust informal mathematical proofs, while many mathematicians would not trust computer proofs. I seemed obvious to me that, on large scale, proving is not only hard but also is imperfect and has engineering character. We need informal proofs and computer proofs and more, e.g. stratification, experimentation.]]></description>
<dc:subject>papers have-read computation constructivism mathematics logic computer-science</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:52efdc87eec1/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:have-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:constructivism"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:mathematics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:logic"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2002.03145">
    <title>[2002.03145] Means-fit effectivity</title>
    <dc:date>2022-07-26T21:04:27+00:00</dc:date>
    <link>https://arxiv.org/abs/2002.03145</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Historically, the notion of effective algorithm is closely related to the Church-Turing thesis. But effectivity imposes no restriction on computation time or any other resource; in that sense, it is incompatible with engineering or physics. We propose a natural generalization of it, means-fitting effectivity, which is effectivity relative to the (physical or abstract) underlying machinery of the algorithm. This machinery varies from one class of algorithms to another. Think for example of ruler-and-compass algorithms, arithmetical algorithms, and Blum-Shub-Smale algorithms. We believe that means-fitting effectivity is meaningful and useful independently of the Church-Turing thesis. Means-fitting effectivity is definable, at least in the theory of abstract state machines (ASMs). The definition elucidates original effectivity as well. Familiarity with the ASM theory is not assumed. We tried to make the paper self-contained. ]]></description>
<dc:subject>papers to-read computation computing-machines complexity</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:01137309b13c/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computing-machines"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/1604.04295">
    <title>[1604.04295] Axiomatizing Analog Algorithms</title>
    <dc:date>2021-07-14T17:24:37+00:00</dc:date>
    <link>https://arxiv.org/abs/1604.04295</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[    We propose a formalization of generic algorithms that includes analog algorithms. This is achieved by reformulating and extending the framework of abstract state machines to include continuous-time models of computation. We prove that every hybrid algorithm satisfying some reasonable postulates may be expressed precisely by a program in a simple and expressive language. 

]]></description>
<dc:subject>papers to-read computation computing-machines analog-computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:0ac3843736fd/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computing-machines"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:analog-computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://arxiv.org/abs/2012.05965">
    <title>[2012.05965] Analog Computation and Representation</title>
    <dc:date>2021-05-03T22:24:44+00:00</dc:date>
    <link>https://arxiv.org/abs/2012.05965</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Relative to digital computation, analog computation has been neglected in the philosophical literature. To the extent that attention has been paid to analog computation, it has been misunderstood. The received view -- that analog computation has to do essentially with continuity -- is simply wrong, as shown by careful attention to historical examples of discontinuous, discrete analog computers. Instead of the received view, I develop an account of analog computation in terms of a particular type of analog representation that allows for discontinuity. This account thus characterizes all types of analog computation, whether continuous or discrete. Furthermore, the structure of this account can be generalized to other types of computation: analog computation essentially involves analog representation, whereas digital computation essentially involves digital representation. Besides being a necessary component of a complete philosophical understanding of computation in general, understanding analog computation is important for computational explanation in contemporary neuroscience and cognitive science. ]]></description>
<dc:subject>papers have-read computation philosophy-of-science</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:9284343b5b03/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:have-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy-of-science"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://deontologistics.co/2019/11/01/tfe-information-and-energy/">
    <title>TfE: Information and Energy – DEONTOLOGISTICS</title>
    <dc:date>2021-05-03T21:48:52+00:00</dc:date>
    <link>https://deontologistics.co/2019/11/01/tfe-information-and-energy/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>information-theory computation cybernetics</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:280635a607f3/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cybernetics"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.sciencedirect.com/science/article/pii/0890540191900524">
    <title>Notions of computation and monads - ScienceDirect</title>
    <dc:date>2017-06-06T20:56:06+00:00</dc:date>
    <link>http://www.sciencedirect.com/science/article/pii/0890540191900524</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>papers to-read categories functional-programming computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:23ba26bba83e/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:categories"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:functional-programming"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://www.microsoft.com/en-us/research/publication/proving-programs-robust/">
    <title>Proving Programs Robust - Microsoft Research</title>
    <dc:date>2017-04-17T16:52:03+00:00</dc:date>
    <link>https://www.microsoft.com/en-us/research/publication/proving-programs-robust/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>papers to-read robustness programming-languages computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:3c344bbc172f/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:robustness"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:programming-languages"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://pages.saclay.inria.fr/olivier.temam/files/eval/DLCPTW2014.pdf">
    <title>Leveraging the Error Resilience of Machine-Learning Applications for Designing Highly Energy Efficient Accelerators</title>
    <dc:date>2015-06-28T04:00:59+00:00</dc:date>
    <link>http://pages.saclay.inria.fr/olivier.temam/files/eval/DLCPTW2014.pdf</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>papers to-read machine-learning hardware computation systems</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:a21d08fbb740/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:machine-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:hardware"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:systems"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://projecteuclid.org/euclid.bj/1377612856">
    <title>Jordan : On statistics, computation and scalability</title>
    <dc:date>2014-12-08T00:03:53+00:00</dc:date>
    <link>http://projecteuclid.org/euclid.bj/1377612856</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[How should statistical procedures be designed so as to be scalable computationally to the massive datasets that are increasingly the norm? When coupled with the requirement that an answer to an inferential question be delivered within a certain time budget, this question has significant repercussions for the field of statistics. With the goal of identifying “time-data tradeoffs,” we investigate some of the statistical consequences of computational perspectives on scability, in particular divide-and-conquer methodology and hierarchies of convex relaxations.]]></description>
<dc:subject>papers to-read statistics big-data computation complexity optimization</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:ddda9ce6c2f6/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:statistics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:big-data"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:optimization"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.oracle.com/us/corporate/features/sparc-m7/index.html">
    <title>SPARC M7 Processor Innovation Leverages Software In Silicon | Oracle</title>
    <dc:date>2014-11-27T05:49:14+00:00</dc:date>
    <link>http://www.oracle.com/us/corporate/features/sparc-m7/index.html</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA["The M7's most significant innovations revolve around what is known as "software in silicon," a design approach that places software functions directly into the processor. Because specific functions are performed in hardware, a software application runs much faster. And because the cores of the processor are freed up to perform other functions, overall operations are speeded up as well."]]></description>
<dc:subject>hardware computation databases</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:85a30f23b848/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:hardware"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:databases"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="https://satana.math.illinois.edu/">
    <title>SATANA</title>
    <dc:date>2013-10-05T19:01:52+00:00</dc:date>
    <link>https://satana.math.illinois.edu/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>UIUC mathematics computation computational-topology</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:ddb97849eee2/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:UIUC"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:mathematics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computational-topology"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://metamarkets.com/2011/machine-learning-in-wonderland/">
    <title>Why Generic Machine Learning Fails</title>
    <dc:date>2012-03-10T02:17:47+00:00</dc:date>
    <link>http://metamarkets.com/2011/machine-learning-in-wonderland/</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>blogs machine-learning online-learning algorithms prediction computation</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:e27184963641/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:blogs"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:machine-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:online-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:prediction"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.podc.org/dijkstra/">
    <title>Edsger W. Dijkstra Prize in Distributed Computing</title>
    <dc:date>2011-11-17T04:53:46+00:00</dc:date>
    <link>http://www.podc.org/dijkstra/</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[The Edsger W. Dijkstra Prize in Distributed Computing is named for Edsger Wybe Dijkstra (1930-2002), a pioneer in the area of distributed computing. His foundational work on concurrency primitives (such as the semaphore), concurrency problems (such as mutual exclusion and deadlock), reasoning about concurrent systems, and self-stabilization comprises one of the most important supports upon which the field of distributed computing is built. No other individual has had a larger influence on research in principles of distributed computing.]]></description>
<dc:subject>computation distributed-computing distributed-systems reference</dc:subject>
<dc:source>https://pinboard.in/</dc:source>
<dc:identifier>https://pinboard.in/u:mraginsky/b:06b7d39d0f56/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:distributed-computing"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:distributed-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:reference"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.cogsci.uci.edu/~ddhoff/ompref.html">
    <title>Observer Mechanics: A Formal Theory of Perception (Bennett, Hoffman, Prakash)</title>
    <dc:date>2011-01-01T19:49:25+00:00</dc:date>
    <link>http://www.cogsci.uci.edu/~ddhoff/ompref.html</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA["Observer Mechanics is an inquiry into the subject of perception. It suggests an approach to the study of perception that attempts to be both rigorous and general. A central thesis of Observer Mechanics is that every perceptual capacity (e.g., stereovision, auditory localization, sentence parsing, haptic recognition, and so on) can be described as an instance of a single formal structure: viz., an "observer.""
]]></description>
<dc:subject>books to-read complexity computation perception dynamical-systems probability multiagent-systems cognitive-science cybernetics</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:9139cac25623/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:books"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:perception"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:dynamical-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:probability"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:multiagent-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cognitive-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:cybernetics"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://gilkalai.wordpress.com/2010/11/09/subexponential-lower-bound-for-randomized-pivot-rules/">
    <title>Subexponential Lower Bound for Randomized Pivot Rules! | Combinatorics and more</title>
    <dc:date>2010-11-10T00:52:21+00:00</dc:date>
    <link>http://gilkalai.wordpress.com/2010/11/09/subexponential-lower-bound-for-randomized-pivot-rules/</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA["Oliver Friedman, Thomas Dueholm Hansen, and Uri Zwick have managed to prove subexponential lower bounds of the form  for ... two basic randomized pivot rules for the simplex algorithm! This is the first result of its kind and deciding if this is possible was an open problem for several decades." And they do it using MDPs!
]]></description>
<dc:subject>papers to-read computer-science computation optimization linear-programming dynamic-programming Markov-decision-processes lower-bounds</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:d6a34feb3e04/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:optimization"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:linear-programming"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:dynamic-programming"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:Markov-decision-processes"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:lower-bounds"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://johncarlosbaez.wordpress.com/2010/10/12/algorithmic-thermodynamics/">
    <title>Algorithmic Thermodynamics « Azimuth</title>
    <dc:date>2010-10-15T18:33:34+00:00</dc:date>
    <link>http://johncarlosbaez.wordpress.com/2010/10/12/algorithmic-thermodynamics/</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[John Baez blogs about his paper with Mike Stay on algorithmic thermodynamics
]]></description>
<dc:subject>have-read blogs information-theory complexity computation thermodynamics</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:f637e58493f9/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:have-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:blogs"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:thermodynamics"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://arxiv.org/abs/1010.2067">
    <title>[1010.2067] Algorithmic Thermodynamics</title>
    <dc:date>2010-10-15T18:32:45+00:00</dc:date>
    <link>http://arxiv.org/abs/1010.2067</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>papers to-read information-theory complexity computation thermodynamics</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:a9660acd8198/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:thermodynamics"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://arxiv.org/abs/1007.5354">
    <title>[1007.5354] Synchronization and Control in Intrinsic and Designed Computation: An Information-Theoretic Analysis of Competing Models of Stochastic Computation</title>
    <dc:date>2010-08-02T00:14:47+00:00</dc:date>
    <link>http://arxiv.org/abs/1007.5354</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA["We adapt tools from information theory to analyze how an observer comes to synchronize with the hidden states of a finitary, stationary stochastic process. We show that synchronization is determined by both the process's internal organization and by an observer's model of it. We analyze these components using the convergence of state-block and block-state entropies, comparing them to the previously known convergence properties of the Shannon block entropy. Along the way, we introduce a hierarchy of information quantifiers as derivatives and integrals of these entropies, which parallels a similar hierarchy introduced for block entropy. We also draw out the duality between synchronization properties and a process's controllability. The tools lead to a new classification of a process's alternative representations in terms of minimality, synchronizability, and unifilarity."
]]></description>
<dc:subject>papers to-read information-theory control-theory complexity computation</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:e1e9eb510b59/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:control-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://arxiv.org/abs/0901.2735">
    <title>[0901.2735] State Space Realization Theorems For Data Mining</title>
    <dc:date>2010-05-12T04:07:55+00:00</dc:date>
    <link>http://arxiv.org/abs/0901.2735</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA["In this paper, we consider formal series associated with events, profiles derived from events, and statistical models that make predictions about events. We prove theorems about realizations for these formal series using the language and tools of Hopf algebras."
]]></description>
<dc:subject>papers to-read state-space-models machine-learning computation</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:573aad27bb6b/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:to-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:state-space-models"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:machine-learning"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.eecs.berkeley.edu/~christos/evol/compevol.htm">
    <title>Computational Aspects of Evolution</title>
    <dc:date>2010-01-24T15:23:37+00:00</dc:date>
    <link>http://www.eecs.berkeley.edu/~christos/evol/compevol.htm</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[A course taught by Christos Papadimitriou
]]></description>
<dc:subject>evolution complexity computation computer-science optimization lecture-notes algorithms</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:fd9dd4e3aecb/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:evolution"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:optimization"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:lecture-notes"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://modelingwithdata.org/arch/00000032.htm">
    <title>The statistics style report (Ben Klemens)</title>
    <dc:date>2009-12-24T21:15:29+00:00</dc:date>
    <link>http://modelingwithdata.org/arch/00000032.htm</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>statistics complexity computation essays philosophy-of-science via:arsyed</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:4f86516e35bb/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:statistics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:essays"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy-of-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:via:arsyed"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.icsi.berkeley.edu/cgi-bin/pubs/browse.pl?groupid=000006">
    <title>International Computer Science Institute | Publications: 2009</title>
    <dc:date>2009-10-09T02:40:29+00:00</dc:date>
    <link>http://www.icsi.berkeley.edu/cgi-bin/pubs/browse.pl?groupid=000006</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[ICSI tech repots
]]></description>
<dc:subject>papers complexity computer-science computation algorithms AI statistics</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:f54c33ffdcb7/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:AI"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:statistics"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://people.csail.mit.edu/madhu/papers/gjs-full.pdf">
    <title>Goldreich, Juba, and Sudan &quot;A Theory of Goal-Oriented Communication&quot;</title>
    <dc:date>2009-10-09T01:00:36+00:00</dc:date>
    <link>http://people.csail.mit.edu/madhu/papers/gjs-full.pdf</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>via:arthegall papers have-read distributed-systems computer-science complexity computation communication filetype:pdf media:document</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:33a9f3b91b94/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:via:arthegall"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:papers"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:have-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:distributed-systems"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:communication"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:filetype:pdf"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:media:document"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://terrytao.wordpress.com/2009/08/05/mosers-entropy-compression-argument/">
    <title>Moser’s entropy compression argument</title>
    <dc:date>2009-08-11T16:47:31+00:00</dc:date>
    <link>http://terrytao.wordpress.com/2009/08/05/mosers-entropy-compression-argument/</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA[Terry Tao explains the recent paper by Robin Moser.
]]></description>
<dc:subject>probability information-theory complexity computer-science mathematics computation</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:530f864d560b/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:probability"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:mathematics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://hunch.net/?p=727">
    <title>Machine Learning (Theory) » Computability in Artificial Intelligence</title>
    <dc:date>2009-05-09T03:58:13+00:00</dc:date>
    <link>http://hunch.net/?p=727</link>
    <dc:creator>mraginsky</dc:creator><description><![CDATA["Let me show by analogy why limiting research to computational questions is bad for any field. Except in computer science, computational aspects play little role in the development of fundamental theories: Consider e.g. set theory with axiom of choice, foundations of logic, exact/full minimax for zero-sum games, quantum (field) theory, string theory, … Indeed, at least in physics, every new fundamental theory seems to be less computable than previous ones."
]]></description>
<dc:subject>blogs have-read science complexity computer-science computation philosophy epistemology AI learning-theory</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:26ed42cbf162/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:blogs"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:have-read"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computer-science"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:philosophy"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:epistemology"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:AI"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:learning-theory"/>
</rdf:Bag></taxo:topics>
</item>
<item rdf:about="http://www.stanford.edu/~montanar/BOOK/book.html">
    <title>http://www.stanford.edu/~montanar/BOOK/book.html</title>
    <dc:date>2009-01-07T00:09:50+00:00</dc:date>
    <link>http://www.stanford.edu/~montanar/BOOK/book.html</link>
    <dc:creator>mraginsky</dc:creator><dc:subject>books statistical-physics information-theory complexity computation algorithms optimization</dc:subject>
<dc:identifier>https://pinboard.in/u:mraginsky/b:83d825e37db7/</dc:identifier>
<taxo:topics><rdf:Bag>	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:books"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:statistical-physics"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:information-theory"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:complexity"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:computation"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:algorithms"/>
	<rdf:li rdf:resource="https://pinboard.in/u:mraginsky/t:optimization"/>
</rdf:Bag></taxo:topics>
</item>
</rdf:RDF>