The idea is interesting but I wonder what is the relationship between "hierarchical/compositional structures" and the manifold hypothesis? https://t.co/0aCnzgFj2q
neural networks approximates complex functions with simpler ones, not 4 all possible functions but 4 only a subset! #DeepLearning favors certain classes of exceptionally simple probability distributions https://t.co/rJlbLUqlHV https://t.co/FQonX5AtX0
Max Tegmark working on deep learning: "Why does deep and cheap learning work so well?" https://t.co/XCG0xSjtdy
Have MIT & Harvard physicists just proved heuristics & approximation are best ways to tackle complexity of universe? https://t.co/EKanhoNBo2
@jeremyjkun This Taylor approximates a small sigmoid NN and shows it can learn XOR (https://t.co/JNShO7MCuh). Doesn't answer why RELUs don't work.
Why neural networks work so well can't be answered with mathematics alone, because part of the answer lies in physics. Max Tegmark, Life 3.0 #NeuralNetworks #AI https://t.co/sxuSrznVjq
Love this paper ... "Why Does Deep Learning and Cheap Learning Work So Well?" https://t.co/T86WGwZfxO #DeepLearning
In anthropics there are not only lower bounds on the complexity of physics needed to sustain the mechanics of intelligent life, but upper bounds on the complexity of the environment generated, since statistical learning relies on powerful compressions a la
@deliprao @dennybritz In page 11 of https://t.co/Ux4ZWJdI0Z they also discuss and look at small experiments suggesting that while depth of linear layers might not have differing expressive powers, they do have different search difficulties wrt representati
Why does deep and cheap learning work so well? https://t.co/wOnuuae0TF こんなんありましたね
This is why many good data scientists are good physicists https://t.co/E82qeqlJQB
これざっと読んでみたけど,結構面白い.輪講で精読したい.https://t.co/rtS8qHRLV3
@_vade @Drokeby @tegmark unrelated but related, @tegmark et al have nice perspective on why #DeepLearning works as well it does: because #physics! (& properties of the natural world) https://t.co/aYI3Ajo9q8
RT @hillbig: 深層学習がうまくいくのは、自然界の 対象の問題の特徴に低次元、局所影響性、対称性がみられ、データ生成過程にマルコフ性が成り立ち、逆向きに推論できるから。自然にみられるデータの特徴付けに物理の考えで切り込んだ物理界20歳すごい。 https://t.co…
@rahuldave I have some thoughts about *what* the implicit prior over functions encoded by neural networks is https://t.co/eC9NykF4n6 ... but less understanding of why this is the right prior for problems we care about. For an interesting perspective, maybe
@ProfPaulStewart not exactly #groundhogday, check out the paper that the article summarizes - https://t.co/KxuAcV4g2i
RT @Roger_M_Taylor: RT @Panarin_Kirill: Interesting view by physicists on why deep learning is actually works and works well in practice #D…
RT @Panarin_Kirill: Interesting view by physicists on why deep learning is actually works and works well in practice #DeepLearning #NeuralNetworks #physics #fb https://t.co/iU4EtySx5B https://t.co/FxhnYc6SHV
RT @Panarin_Kirill: Interesting view by physicists on why deep learning is actually works and works well in practice #DeepLearning #NeuralN…
RT @Panarin_Kirill: Interesting view by physicists on why deep learning is actually works and works well in practice #DeepLearning #NeuralN…
RT @Panarin_Kirill: Interesting view by physicists on why deep learning is actually works and works well in practice #DeepLearning #NeuralN…
Interesting view by physicists on why deep learning is actually works and works well in practice #DeepLearning #NeuralNetworks #physics #fb https://t.co/hO7TeOlQxh
Nobody understands why deep neural networks are so good at solving complex problems. Now physicists say the... https://t.co/fQcbgZGSyV
Why does deep and cheap learning work so well? https://t.co/ZvNV3VkmW7
Why deep learning works so well. https://t.co/g1Um3FqPa6
@yuval_v @zdanielz אוקיי יש קשר עקיף כי סל׳ אוטו׳ קשורים למכניקה סטטיסטית שיש לה קשרים ידועים לDL, כמו המאמר של טגמארק על רהנורמליזציה. https://t.co/thrVWf0pii
So why #deeplearning works can be explained by #physicist not #mathematician ? Interesting. https://t.co/0VMjoodDhd
"Why does deep and cheap learning work so well?" Lin, @tegmark, Rolnick https://t.co/BHXq6AenMt https://t.co/ZTg9SGum6u
Por que as redes neurais profundas são tão efetivas? Talvez porque explorem a estrutura fundamental da Realidade. https://t.co/di8qOjHNbw https://t.co/nuhlucBAI2
Ref: https://t.co/SrRNcEYoa5: Why Does Deep and Cheap Learning Work So Well? https://t.co/mz3UaTam0c
RT @LauriLoven: Holy Crap. "Why does deep and cheap learning work so well? We show how the success of deep learning depends... https://t…
RT @sramji: "Why does deep and cheap learning work so well?" a mathematical exploration by @tegmark + Henry Lin + David Rolnick https://t.c…
Max Tegmark relates deep learning to renormalization group. Why does deep learning work so well?: https://t.co/onyyHIscrN #ITRTG
"Why does deep and cheap learning work so well?" a mathematical exploration by @tegmark + Henry Lin + David Rolnick https://t.co/BZyaDuronR
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
[1608.08225] Why does deep and cheap learning work so well? https://t.co/Bl1Cqx2l3T 一年前の話題になったこの論文,4訂されてますね…
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
#arXiv #stat_ML "Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED)" https://t.co/0lGLStUKWV
Why does deep and cheap learning work so well?. (arXiv:1608.08225v4 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
@tdietterich @deliprao One of the best refs along these lines is Lin/Tegmark's (speculative) argument about actual data vs. possible data: https://t.co/FQG9VZrw9a
Why does cheap and Deep learning work so well?https://t.co/uCW3jCPud6 <Why is the universe explained with low order polynomial Hamiltonians?
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
RT @gp_pulipaka: In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1o…
In What Sense Does #DeepLearning Reflect the Laws of #Physics? #BigData #MachineLearning #DataScience #AI https://t.co/1oezbZFaDF https://t.co/Ov1rv6Kl5C
ICYMI Just why does deep learning work? https://t.co/bNZW6YUOTn
very curious ! Describe Deep Learning by the laws of physics https://t.co/1CUir6gKpq
RT @latone: cosmologists explain #deeplearning, "Why does deep and cheap learning work so well?" by @tegmark, et al, https://t.co/BHXq6AenMt
cosmologists explain #deeplearning, "Why does deep and cheap learning work so well?" by @tegmark, et al, https://t.co/BHXq6AenMt
.@neiltyson has made mistakes wading into biology before, but I contend this is computer science & v interesting. EG https://t.co/Ko87aovShe https://t.co/QvKKdmPTPy
なんだこのユーモアあふれるタイトルは https://t.co/JtR1hlbv65
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v3 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
Why does deep and cheap learning work so well?. (arXiv:1608.08225v3 [cond-mat.dis-nn] UPDATED) https://t.co/EDSAsRH0Fe
RT @StatMLPapers: Why does deep and cheap learning work so well?. (arXiv:1608.08225v3 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
https://t.co/1MwATos06P こういうことしたい。
Why does deep and cheap learning work so well?. (arXiv:1608.08225v3 [cond-mat.dis-nn] UPDATED) https://t.co/sJvjdeYTfo
#arXiv #stat_ML "Why does deep and cheap learning work so well?. (arXiv:1608.08225v3 [cond-mat.dis-nn] UPDATED)" https://t.co/0lGLStUKWV
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning... https://t.co/KQ4lTqS6LI by #alevergara78 https://t.co/e6bfbB0utY
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…
RT @gp_pulipaka: Why does deep and cheap learning work so well? #BigData #DeepLearning #MachineLearning #DataScience #AI https://t.co/MPoU…