PPT-Provable Learning of Noisy-OR Networks

Rong Ge Duke University Joint work with Sanjeev Arora Tengyu Ma Andrej Risteski Provable Learning of NoisyOR Networks STOC 2017 arxiv161208795 New practical algorithms for learning NoisyOR networks via symmetric NMF. Models, Adversaries, Reductions. Cryptography / Cryptology. “from.  . Greek.  . κρυ. πτός.  . kryptós. , "hidden, secret"; and . γράφειν. . graphein. , "writing", or . -λογία. Cost function. Machine Learning. Neural Network (Classification). Binary classification. . . 1 output unit. Layer 1. Layer 2. Layer 3. Layer 4. Multi-class classification . (K classes). K output units. Deep Learning @ . UvA. UVA Deep Learning COURSE - Efstratios Gavves & Max Welling. LEARNING WITH NEURAL NETWORKS . - . PAGE . 1. Machine Learning Paradigm for Neural Networks. The Backpropagation algorithm for learning with a neural network. Chun Lam Chan. , Pak . Hou. . Che. and . Sidharth. . Jaggi. The Chinese University of Hong Kong. Venkatesh. . Saligrama. Boston University. Non-adaptive probabilistic group testing with noisy measurements: Near-optimal bounds with efficient algorithms. A science story. Darin J. . Ulness. Department of Chemistry. Concordia College, Moorhead, MN. Spectroscopy. Using . light. to gain information about . matter. Spectra. Transition frequencies. Time dynamics. . Defaults. Foundations of AI. Given an N X N matrix, and given N colors, color the matrix in such a way that:. -all cells are colored;. - each color occurs exactly once in each row;. - each color occurs exactly once in each column;. Dani Koleva and George Bogdanov. Inspired by Richard Bennett in March 2010. 30 April 2014, Sofia, Bulgaria. 1. NETWORKS. Why?. When?. What?. but mainly…. How?. 2. WHY NETWORKS?. 3. WHY NETWORKS?. Learning from/with each other: becoming excellent at what we do. . Defaults. Foundations of AI. Given an N X N matrix, and given N colors, color the matrix in such a way that:. -all cells are colored;. - each color occurs exactly once in each row;. - each color occurs exactly once in each column;. . Rob Fergus (New York University). Yair Weiss (Hebrew University). Antonio Torralba (MIT). . Presented by Gunnar Atli Sigurdsson. TexPoint fonts used in EMF. . Read the TexPoint manual before you delete this box.: AAAAAAAAAA. . PAGE 138 to 142. . fact. An objective statement, one that can be proved . opinion. One possible interpretation of the facts, subjective statement that is not based on proof . Objective. Provable . Secada combs | bus-550. AI Superpowers: china, silicon valley, and the new world order. Kai Fu Lee. Author of AI Superpowers. Currently Chairman and CEO of . Sinovation. Ventures and President of . Sinovation. blatant . (adj). noisy in a coarse, offensive way. obvious or conspicuous, especially in an unfavorable way. ---------------------------------. inconsequential. broach. (v/n). to bring up or begin to talk about (a subject). Spikes in trigger rate. Periodic:. With B ON in 2008 . Without B on during MWGR18 . Sporadic . MWGR 19. Strip noise profile. 6 may . 22 April. REASON: HV problem in RB1 out sect 12. Noisy topology. Jiang. Feb 17. Model formulation.  .  .  .  .  .  . …. Recall the model of fully-connected neural networks.  . When .  . Linear Networks. In the following slides, we only consider linear networks without bias:.