PPT-Multimodal Learning with Deep Boltzmann Machines

Author Nitish Srivastava Ruslan Salakhutdinov Presenter Shuochao Yao Data Collection of Modalities Multimedia content on the web image text audio Product recommendation systems. stanfordedu Aditya Khosla aditya86csstanfordedu Mingyu Kim minkyu89csstanfordedu Juhan Nam juhanccrmastanfordedu Honglak Lee honglakeecsumichedu Andrew Y Ng angcsstanfordedu Computer Science Department Stanford University Stanf torontoedu Geoffrey Hinton Department of Computer Science University of Toronto hintoncstorontoedu Abstract We present a new learning algorithm for Boltz mann machines that contain many layers of hid den variables Datadependent expectations are estim torontoedu Ruslan Salakhutdinov Department of Statistics and Computer Science University of Toronto rsalakhucstorontoedu Abstract A Deep Boltzmann Machine is described for learning a generative model of data that consists of multiple and diverse inpu stanfordedu Aditya Khosla aditya86csstanfordedu Mingyu Kim minkyu89csstanfordedu Juhan Nam juhanccrmastanfordedu Honglak Lee honglakeecsumichedu Andrew Y Ng angcsstanfordedu Computer Science Department Stanford University Stanf Restricted Boltzmann machines RBMs are probabilistic graphical models that can be interpreted as stochastic neural networks Theincreaseincomputationalpowerandthedevelopmentoffasterlearn ing algorithms have made them applicable to relevant machine le Ng Computer Science Department Stanford University jngiamaditya86minkyu89ang csstanfordedu Department of Music Stanford University juhanccrmastanfordedu Computer Science Engineering Division University of Michigan Ann Arbor honglakeecsumichedu Abst Ng Computer Science Department Stanford University jngiamaditya86minkyu89ang csstanfordedu Department of Music Stanford University juhanccrmastanfordedu Computer Science Engineering Division University of Michigan Ann Arbor honglakeecsumichedu Abst torontoedu Geoffrey Hinton Department of Computer Science University of Toronto hintoncstorontoedu Abstract We present a new learning algorithm for Boltz mann machines that contain many layers of hid den variables Datadependent expectations are estim , Ashish Kapoor and Krysta Svore. Microsoft Research. ASCR Workshop. Washington DC. 1412.3489. Quantum Deep Learning. How do we make computers that . see. , . listen. , and . understand. ?. Goal. : Learn . 1. Boltzmann Machine. Relaxation net with visible and hidden units. Learning algorithm. Avoids local minima (and speeds up learning) by using simulated annealing with stochastic nodes. Node activation: Logistic Function. Aaron Crandall, 2015. What is Deep Learning?. Architectures with more mathematical . transformations from source to target. Sparse representations. Stacking based learning . approaches. Mor. e focus on handling unlabeled data. Yulia Kogan and . Ron . Shiff. 19.06.2016. References. J. Mao, W. Xu, Y. Yang, J. Wang, and A. L. Yuille. Explain images with multimodal recurrent neural networks. . arXiv preprint arXiv:1410.1090, 2014. Fall 2018/19. 9. Hopfield Networks, Boltzmann Machines. . Unsupervised Neural Networks. Noriko Tomuro. 2. Hopfield Networks. Concepts. Boltzmann Machines. Concepts. Restricted Boltzmann Machines. Deep Boltzmann Machines. Week 7 Video 3. Thank you. Thank you to . Yiqiu. (Rachel) Zou for feedback and comments on this video. Multimodal Learning Analytics. “A set of techniques that can be used to collect multiple sources of data in high-frequency (video, logs, audio, gestures, biosensors), synchronize and code the data, and examine learning in realistic, ecologically valid, social, mixed-media learning environments.” (.