DOCSLIB.ORG

Deep Learning I

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Deep Learning I Slides adapted from Vlad Morariu CMSC 422 Introduction to Machine Learning Lecture 20 Deep Learning I Furong Huang / [email protected] History of Neural Networks 2 Standard computer vision pipeline “cat” or “background” classificatio n feature extraction features predicted labels “cat” Features: HOG, SIFT, LBP, … supervision training Classifiers: SVM, RF, KNN, … Features are hand-crafted, not trained eventually limited by feature quality Cat image credit: https://raw.githubusercontent.com/BVLC/caffe/master/examples/images/cat.jpg 3 Deep learning training supervision features classifier Image credit: LeCun, Y., Bottou, L., Bengio, Y., Haffner, P. “Gradient-based learning applied to document recognition.” Proceedings of the IEEE, 1998. Deep learning multiple layer neural networks learn features and classifiers directly (“end-to-end” training) 4 SWITCHBOARD: telephone speech corpus for research and development Speech Recognition Slide credit: Bohyung Han 5 Image Classification Performance Image Classification Top-5 Errors (%) Figure from: K. He, X. Zhang, S. Ren, J. Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015. (slides) 6 Slide credit: Bohyung Han Biological inspiration Image source: http://cs231n.github.io/neural-networks-1/ 7 Artificial neuron Image source: http://cs231n.github.io/neural-networks-1/ Activation function is usually non-linear step, tanh, sigmoid The actual biological system is much more complicated 8 McCulloch-Pitts Model Read: https://en.wikipedia.org/wiki/Artificial_neuron • Threshold Logic Unit (TLU) ➢ Warren McCulloch and Walter Pitts, 1943 ➢ Binary inputs/outputs and Threshold activation function ➢ Can represent AND/OR/NOT functions which can be composed for complex functions 9 Rosenblatt’s Perceptron Read: https://en.wikipedia.org/wiki/Perceptron • Perceptron ➢ Proposed by Frank Rosenblatt in 1957 ➢ Based on McCulloch-Pitts model ➢ Real inputs/outputs, threshold activation function 10 Perceptron Learning • Given a training dataset of input features and labels • Initialize weights randomly • For each example in training set ➢ Classify example using current weights ➢ Update weights • If data is linearly separable, convergence is guaranteed in a bounded number of iterations https://en.wikipedia.org/wiki/Perceptron 11 Multiple output variables Weights to predict multiple outputs can be learned independently b1 x1 w11 w12 y1 w13 w14 x 2 b2 w21 w22 w 23 y2 w24 x3 b w31 3 w32 w33 y3 x4 w34 inputs outputs 12 Perceptron success • Implemented as custom-built hardware, the “Mark I Perceptron” • Input: photocells • Weights: potentiometers • Weight updates: electric motors • Demonstrated ability to classify 20x20 images • Generated lots of AI excitement • In 1958, the New York Times reported the perceptron to be • "the embryo of an electronic computer that [the Navy] expects will be able to walk, talk, see, write, reproduce itself and be conscious of its existence." https://en.wikipedia.org/wiki/Perceptron 13 Linear Classifier Slide credit: Bohyung Han 14 Nonlinear Classifier Slide credit: Bohyung Han 15 Nonlinear Classifier – XOR problem 16 Perceptron limitations • If there are multiple separating hyperplanes, learning will converge to one of them (not the optimal one) • If training set is not linearly separable, training will fail completely • Marvin Minsky and Seymour Papert, “Perceptrons”, 1969 • Proved that it was impossible to learn an XOR function with a single layer perceptron network • Led to the “AI Winter” of the 1970’s https://en.wikipedia.org/wiki/Perceptron 17 Multi-Layer Perceptron (MLP) Image source: http://cs231n.github.io/neural-networks-1/ • Activation function need not be a threshold • Multiple layers can represent XOR function • But perceptron algorithm cannot be used to update weights • Why? Hidden layers are not observed! https://en.wikipedia.org/wiki/Multilayer_perceptron 18 Multi-Layer: Backpropagation 푦 푤푘푖 푧푖 푤푖푗 푧푗 푥푘 횺 푦ො푖 횺 푦ො푗 퓛 퐸 Sigmoid Sigmoid Neuron 푖 Neuron 푗 휕퐸 휕퐸 푑푦ො푗 = 휕푧푗 휕푦ො푗 푑푧푗 휕퐸 휕퐸 푑푧푗 휕퐸 휕퐸 푑푦ො푗 = ෍ = ෍ 푤푖푗 = ෍ 푤푖푗 휕푦ො푖 휕푧푗 푑푦ො푖 휕푧푗 휕푦ො푗 푑푧푗 푗 푗 푗 푛 푛 푛 푛 푛 휕퐸 휕퐸 푑푦ො푖 휕푧푖 푑푦ො푖 휕푧푖 휕퐸 푑푦ො푗 = ෍ 푛 푛 = ෍ 푛 ෍ 푤푖푗 푛 푛 휕푤 휕푦ො푖 푑푧푖 휕푤푘푖 푑푧 휕푤푘푖 휕푦ො 푑푧 푘푖 푛 푛 푖 푗 푗 푗 Slide credit: Bohyung Han 19 Stochastic Gradient Descent (SGD) Update weights for each sample 푛 1 푛 푛 2 휕퐸 퐸 = 푦 − 푦ො 풘푖 푡 + 1 = 풘푖 푡 − 휖 2 휕풘푖 + Fast, online − Sensitive to noise Minibatch SGD: Update weights for a small set of samples 퐵 1 푛 푛 2 휕퐸 퐸 = ෍ 푦 − 푦ො 풘푖 푡 + 1 = 풘푖 푡 − 휖 2 휕풘푖 푛∈퐵 + Fast, online + Robust to noise Slide credit: Bohyung Han 20 Momentum Remember the previous direction + Converge faster + Avoid oscillation 휕퐸 푣푖 푡 = 훼푣푖 푡 − 1 − 휖 (푡) 휕푤푖 풘 푡 + 1 = 풘 푡 + 풗(푡) Slide credit: Bohyung Han 21 Weight Decay Penalize the size of the weights 1 퐶 = 퐸 + ෍ 푤2 2 푖 푖 휕퐶 휕퐸 푤푖 푡 + 1 = 푤푖 푡 − 휖 = 푤푖 푡 − 휖 − 휆푤푖 휕푤푖 휕푤푖 + Improve generalization a lot! Slide credit: Bohyung Han22 Furong Huang 3251 A.V. Williams, College Park, MD 20740 301.405.8010 / [email protected].
Load more

Recommended publications
  • Lecture 6 Learned Feedforward Visual Processing Neural Networks, Deep Learning, Convnets
    William T. Freeman, Antonio Torralba, 2017 Lecture 6 Learned feedforward visual processing Neural Networks, Deep learning, ConvNets Some slides modified from R. Fergus We need translation invariance Lots of useful linear filters… Laplacian Gaussian derivative Gaussian Gabor And many more… High order Gaussian derivatives We need translation and scale invariance Lots of image pyramids… Gaussian Pyr Laplacian Pyr And many more: QMF, steerable, … We need … What is the best representation? • All the previous representation are manually constructed. • Could they be learnt from data? A brief history of Neural Networks enthusiasm time Perceptrons, 1958 Rosenblatt http://www.ecse.rpi.edu/homepages/nagy/PDF_chrono/2011_Na gy_Pace_FR.pdf. Photo by George Nagy 9 http://www.manhattanrarebooks-science.com/rosenblatt.htm Perceptrons, 1958 10 Perceptrons, 1958 enthusiasm time Minsky and Papert, Perceptrons, 1972 12 Perceptrons, 1958 enthusiasm Minsky and Papert, 1972 time Parallel Distributed Processing (PDP), 1986 14 XOR problem Inputs Output 0 0 0 1 0 1 0 1 1 0 1 1 1 0 0 1 PDP authors pointed to the backpropagation algorithm as a breakthrough, allowing multi-layer neural networks to be trained. Among the functions that a multi-layer network can represent but a single-layer network cannot: the XOR function. 15 Perceptrons, PDP book, 1958 1986 enthusiasm Minsky and Papert, 1972 time LeCun conv nets, 1998 Demos: http://yann.lecun.com/exdb/lenet/index.html 17 18 Neural networks to recognize handwritten digits? yes Neural networks for tougher problems? not really http://pub.clement.farabet.net/ecvw09.pdf 19 NIPS 2000 • NIPS, Neural Information Processing Systems, is the premier conference on machine learning.
    [Show full text]
  • Comparative Study of Deep Learning Software Frameworks
    Comparative Study of Deep Learning Software Frameworks Soheil Bahrampour, Naveen Ramakrishnan, Lukas Schott, Mohak Shah Research and Technology Center, Robert Bosch LLC {Soheil.Bahrampour, Naveen.Ramakrishnan, fixed-term.Lukas.Schott, Mohak.Shah}@us.bosch.com ABSTRACT such as dropout and weight decay [2]. As the popular- Deep learning methods have resulted in significant perfor- ity of the deep learning methods have increased over the mance improvements in several application domains and as last few years, several deep learning software frameworks such several software frameworks have been developed to have appeared to enable efficient development and imple- facilitate their implementation. This paper presents a com- mentation of these methods. The list of available frame- parative study of five deep learning frameworks, namely works includes, but is not limited to, Caffe, DeepLearning4J, Caffe, Neon, TensorFlow, Theano, and Torch, on three as- deepmat, Eblearn, Neon, PyLearn, TensorFlow, Theano, pects: extensibility, hardware utilization, and speed. The Torch, etc. Different frameworks try to optimize different as- study is performed on several types of deep learning ar- pects of training or deployment of a deep learning algorithm. chitectures and we evaluate the performance of the above For instance, Caffe emphasises ease of use where standard frameworks when employed on a single machine for both layers can be easily configured without hard-coding while (multi-threaded) CPU and GPU (Nvidia Titan X) settings. Theano provides automatic differentiation capabilities which The speed performance metrics used here include the gradi- facilitates flexibility to modify architecture for research and ent computation time, which is important during the train- development. Several of these frameworks have received ing phase of deep networks, and the forward time, which wide attention from the research community and are well- is important from the deployment perspective of trained developed allowing efficient training of deep networks with networks.
    [Show full text]
  • Comparative Study of Caffe, Neon, Theano, and Torch
    Workshop track - ICLR 2016 COMPARATIVE STUDY OF CAFFE,NEON,THEANO, AND TORCH FOR DEEP LEARNING Soheil Bahrampour, Naveen Ramakrishnan, Lukas Schott, Mohak Shah Bosch Research and Technology Center fSoheil.Bahrampour,Naveen.Ramakrishnan, fixed-term.Lukas.Schott,[email protected] ABSTRACT Deep learning methods have resulted in significant performance improvements in several application domains and as such several software frameworks have been developed to facilitate their implementation. This paper presents a comparative study of four deep learning frameworks, namely Caffe, Neon, Theano, and Torch, on three aspects: extensibility, hardware utilization, and speed. The study is per- formed on several types of deep learning architectures and we evaluate the per- formance of the above frameworks when employed on a single machine for both (multi-threaded) CPU and GPU (Nvidia Titan X) settings. The speed performance metrics used here include the gradient computation time, which is important dur- ing the training phase of deep networks, and the forward time, which is important from the deployment perspective of trained networks. For convolutional networks, we also report how each of these frameworks support various convolutional algo- rithms and their corresponding performance. From our experiments, we observe that Theano and Torch are the most easily extensible frameworks. We observe that Torch is best suited for any deep architecture on CPU, followed by Theano. It also achieves the best performance on the GPU for large convolutional and fully connected networks, followed closely by Neon. Theano achieves the best perfor- mance on GPU for training and deployment of LSTM networks. Finally Caffe is the easiest for evaluating the performance of standard deep architectures.
    [Show full text]
  • Tensorflow, Theano, Keras, Torch, Caffe Vicky Kalogeiton, Stéphane Lathuilière, Pauline Luc, Thomas Lucas, Konstantin Shmelkov Introduction
    TensorFlow, Theano, Keras, Torch, Caffe Vicky Kalogeiton, Stéphane Lathuilière, Pauline Luc, Thomas Lucas, Konstantin Shmelkov Introduction TensorFlow Google Brain, 2015 (rewritten DistBelief) Theano University of Montréal, 2009 Keras François Chollet, 2015 (now at Google) Torch Facebook AI Research, Twitter, Google DeepMind Caffe Berkeley Vision and Learning Center (BVLC), 2013 Outline 1. Introduction of each framework a. TensorFlow b. Theano c. Keras d. Torch e. Caffe 2. Further comparison a. Code + models b. Community and documentation c. Performance d. Model deployment e. Extra features 3. Which framework to choose when ..? Introduction of each framework TensorFlow architecture 1) Low-level core (C++/CUDA) 2) Simple Python API to define the computational graph 3) High-level API (TF-Learn, TF-Slim, soon Keras…) TensorFlow computational graph - auto-differentiation! - easy multi-GPU/multi-node - native C++ multithreading - device-efficient implementation for most ops - whole pipeline in the graph: data loading, preprocessing, prefetching... TensorBoard TensorFlow development + bleeding edge (GitHub yay!) + division in core and contrib => very quick merging of new hotness + a lot of new related API: CRF, BayesFlow, SparseTensor, audio IO, CTC, seq2seq + so it can easily handle images, videos, audio, text... + if you really need a new native op, you can load a dynamic lib - sometimes contrib stuff disappears or moves - recently introduced bells and whistles are barely documented Presentation of Theano: - Maintained by Montréal University group. - Pioneered the use of a computational graph. - General machine learning tool -> Use of Lasagne and Keras. - Very popular in the research community, but not elsewhere. Falling behind. What is it like to start using Theano? - Read tutorials until you no longer can, then keep going.
    [Show full text]
  • DIY Deep Learning for Vision: the Caffe Framework
    DIY Deep Learning for Vision: the Caffe framework caffe.berkeleyvision.org github.com/BVLC/caffe Evan Shelhamer adapted from the Caffe tutorial with Jeff Donahue, Yangqing Jia, and Ross Girshick. Why Deep Learning? The Unreasonable Effectiveness of Deep Features Classes separate in the deep representations and transfer to many tasks. [DeCAF] [Zeiler-Fergus] Why Deep Learning? The Unreasonable Effectiveness of Deep Features Maximal activations of pool5 units [R-CNN] conv5 DeConv visualization Rich visual structure of features deep in hierarchy. [Zeiler-Fergus] Why Deep Learning? The Unreasonable Effectiveness of Deep Features 1st layer filters image patches that strongly activate 1st layer filters [Zeiler-Fergus] What is Deep Learning? Compositional Models Learned End-to-End What is Deep Learning? Compositional Models Learned End-to-End Hierarchy of Representations - vision: pixel, motif, part, object - text: character, word, clause, sentence - speech: audio, band, phone, word concrete abstract learning What is Deep Learning? Compositional Models Learned End-to-End figure credit Yann LeCun, ICML ‘13 tutorial What is Deep Learning? Compositional Models Learned End-to-End Back-propagation: take the gradient of the model layer-by-layer by the chain rule to yield the gradient of all the parameters. figure credit Yann LeCun, ICML ‘13 tutorial What is Deep Learning? Vast space of models! Caffe models are loss-driven: - supervised - unsupervised slide credit Marc’aurelio Ranzato, CVPR ‘14 tutorial. Convolutional Neural Nets (CNNs): 1989 LeNet: a layered model composed of convolution and subsampling operations followed by a holistic representation and ultimately a classifier for handwritten digits. [ LeNet ] Convolutional Nets: 2012 AlexNet: a layered model composed of convolution, + data subsampling, and further operations followed by a holistic + gpu representation and all-in-all a landmark classifier on + non-saturating nonlinearity ILSVRC12.
    [Show full text]
  • Caffe in Practice
    This Business of Brewing: Caffe in Practice caffe.berkeleyvision.org Evan Shelhamer github.com/BVLC/caffe from the tutorial by Evan Shelhamer, Jeff Donahue, Yangqing Jia, and Ross Girshick Deep Learning, as it is executed... What should a framework handle? Compositional Models Decompose the problem and code! End-to-End Learning Solve and check! Vast Space of Architectures and Tasks Define, experiment, and extend! Frameworks ● Torch7 ○ NYU ○ scientific computing framework in Lua ○ supported by Facebook ● Theano/Pylearn2 ○ U. Montreal ○ scientific computing framework in Python ○ symbolic computation and automatic differentiation ● Cuda-Convnet2 ○ Alex Krizhevsky ○ Very fast on state-of-the-art GPUs with Multi-GPU parallelism ○ C++ / CUDA library Framework Comparison ● More alike than different ○ All express deep models ○ All are nicely open-source ○ All include scripting for hacking and prototyping ● No strict winners – experiment and choose the framework that best fits your work ● We like to brew our deep networks with Caffe Why Caffe? In one sip… ● Expression: models + optimizations are plaintext schemas, not code. ● Speed: for state-of-the-art models and massive data. ● Modularity: to extend to new tasks and architectures. ● Openness: common code and reference models for reproducibility. ● Community: joint discussion, development, and modeling. CAFFE EXAMPLES + APPLICATIONS Share a Sip of Brewed Models demo.caffe.berkeleyvision.org demo code open-source and bundled Scene Recognition by MIT Places CNN demo B. Zhou et al. NIPS 14 Object Detection R-CNN: Regions with Convolutional Neural Networks http://nbviewer.ipython.org/github/BVLC/caffe/blob/master/examples/detection.ipynb Full R-CNN scripts available at https://github.com/rbgirshick/rcnn Ross Girshick et al.
    [Show full text]
  • Deep Learning with Caffe in Python – Part I: Defining a Layer in This Lecture, We Will Discuss How to Get Started with Caffe and Use Its Various Features
    Deep Learning With Caffe In Python – Part I: Defining A Layer In this lecture, we will discuss how to get started with Caffe and use its various features. We will then build a convolutional neural network (CNN) that can be used for image classification. Caffe plays very well with the GPU during the training process, hence we can achieve a lot of speed-up. For the purpose of this discussion, it is assumed that you have already installed Caffe on your machine. Let’s go ahead and see how to interact with Caffe, shall we? Prerequisites Create a python file and add the following lines: import sys import numpy as np import matplotlib.pyplot as plt sys.insert('/path/to/caffe/python') import caffe If you have a GPU onboard, then we need to tell Caffe that we want it to use the GPU: caffe.set_device(0) caffe.set_mode_gpu() We are now ready to build a network. Building a simple layer As the name suggests, convolutional neural networks (CNNs) rely heavily on convolutions. Big surprise, right? CNNs are still basically neural networks, which means they consist of multiple layers joined together. There are many different types of layers that can be used to build a CNN, convolution layer being one of them. Let’s go ahead and see how we can define a simple convolution layer in Caffe. Create a file called “myconvnet.prototxt” and add the following lines to it: name: "myconvolution" input: "data" input_dim: 1 input_dim: 1 input_dim: 256 input_dim: 256 layer { name: "conv" type: "Convolution" bottom: "data" top: "conv" convolution_param { num_output: 10 kernel_size: 3 stride: 1 weight_filler { type: "gaussian" std: 0.01 } bias_filler { type: "constant" value: 0 } } } We just defined a single layer CNN consisting of 10 convolutional neurons (as specified by “num_output”) with a kernel size of 3×3 (as specified by “kernel_size”) and a stride of 1 (as specified by “stride”).
    [Show full text]
  • Bigdl, for Apache Spark* Openvino, Ray* and Apache Spark*
    Cluster Serving: Distributed and Automated Model Inference on Big Data Streaming Frameworks Authors: Jiaming Song, Dongjie Shi, Qiyuan Gong, Lei Xia, Jason Dai Outline Challenges AI productions facing Integrated Big Data and AI pipeline Scalable online serving Cross-industry end-to-end use cases Big Data & Model Performance “Machine Learning Yearning”, Andrew Ng, 2016 *Other names and brands may be claimed as the property of others. Real-World ML/DL Applications Are Complex Data Analytics Pipelines “Hidden Technical Debt in Machine Learning Systems”, Sculley et al., Google, NIPS 2015 Paper *Other names and brands may be claimed as the property of others. Outline Challenges AI productions facing Integrated Big Data and AI pipeline Scalable online serving Cross-industry end-to-end use cases Integrated Big Data Analytics and AI Seamless Scaling from Laptop to Distributed Big Data Prototype on laptop Experiment on clusters Production deployment w/ using sample data with history data distributed data pipeline Production Data pipeline • Easily prototype end-to-end pipelines that apply AI models to big data • “Zero” code change from laptop to distributed cluster • Seamlessly deployed on production Hadoop/K8s clusters • Automate the process of applying machine learning to big data *Other names and brands may be claimed as the property of others. AI on Big Data Distributed, High-Performance Unified Analytics + AI Platform Deep Learning Framework for TensorFlow*, PyTorch*, Keras*, BigDL, for Apache Spark* OpenVINO, Ray* and Apache Spark* https://github.com/intel-analytics/bigdl https://github.com/intel-analytics/analytics-zoo Seamless Scaling from Laptop to Distributed Big Data *Other names and brands may be claimed as the property of others.
    [Show full text]
  • Distributed Deep Q-Learning
    Distributed Deep Q-Learning Kevin Chavez1, Hao Yi Ong1, and Augustus Hong1 Abstract— We propose a distributed deep learning model unsupervised image classification. To achieve model paral- to successfully learn control policies directly from high- lelism, we use Caffe, a deep learning framework developed dimensional sensory input using reinforcement learning. The for image recognition that distributes training across multiple model is based on the deep Q-network, a convolutional neural network trained with a variant of Q-learning. Its input is processor cores [11]. raw pixels and its output is a value function estimating The contributions of this paper are twofold. First, we future rewards from taking an action given a system state. develop and implement a software framework adapted that To distribute the deep Q-network training, we adapt the supports model and data parallelism for DQN. Second, we DistBelief software framework to the context of efficiently demonstrate and analyze the performance of our distributed training reinforcement learning agents. As a result, the method is completely asynchronous and scales well with the number RL agent. The rest of this paper is organized as follows. of machines. We demonstrate that the deep Q-network agent, Section II introduces the background on the class of machine receiving only the pixels and the game score as inputs, was able learning problem our algorithm solves. This is followed by to achieve reasonable success on a simple game with minimal Section III and Section IV, which detail the serial DQN parameter tuning. and our approach to distributing the training. Section V I. INTRODUCTION discusses our experiments on a classic video game, and some Reinforcement learning (RL) agents face a tremendous concluding remarks are drawn and future works mentioned challenge in optimizing their control of a system approaching in Section VI.
    [Show full text]
  • DIY Deep Learning for Vision: a Hands-On Tutorial with Caffe
    Deep Neural Networks Basics For ECS 289G Presented by Fanyi Xiao Computer Vision in the Pre-DNN Era Computer Vision in the Pre-DNN Era Computer Vision in the Pre-DNN Era Computer Vision in the Pre-DNN Era Emergence of DNNs in Vision Neural Networks Learn visual features "end-to- end" Neural Networks Compositional Models Learned End-to-End Hierarchy of Representations - vision: pixel, motif, part, object - text: character, word, clause, sentence - speech: audio, band, phone, word concrete abstract learning figure credit Yann LeCun, ICML ‘13 tutorial Neural Networks Three key ingredients for training an NN: 1. Score function 2. Loss function 3. Optimization Neural Networks Three key ingredients for training an NN: 1. Score function: y=f(x,W) x -- 224*224*3 image patch y -- 1000d vector Neural Networks Three key ingredients for training an NN: 2. Loss function: for example max-margin loss and cross-entropy loss Neural Networks Three key ingredients for training an NN: 3. Optimization: simple gradient descent! Neural Networks Three key ingredients for training an NN: 3. Optimization: in practice, stochasitic (mini-batch) gradient descent! Neural Networks Three key ingredients for training an NN: 3. Optimization: in practice, stochasitic (mini-batch) gradient descent + momentum! (Many other optimization methods like adagrad/rmsprop) Convolution Neural Networks Linear transformation: Let's take a closer look at AlexNet y'=Wx+b Convolution Neural Networks Linear transformation: Let's take a closer look at AlexNet y'=f(Wx+b) Convolution Neural
    [Show full text]
  • DIY Deep Learning for Vision: a Guest Lecture with Caffe
    DIY Deep Learning for Vision: a guest lecture with Caffe caffe.berkeleyvision.org github.com/BVLC/caffe Evan Shelhamer UC Berkeley Based on a tutorial by E. Shelhamer, J. Donahue, Y. Jia, and R. Girshick Machine Learning in one slide Supervised f(x) = y, p(y|x) Unsupervised p(x), f(x) = ? figure credit K. Cho Loss and Risk Loss: a loss (error, or cost) function specifies the goal of learning by mapping parameter settings to a scalar value specifying the “badness” of these parameter settings Risk Empirical Risk Risk: risk measures loss over the input. Since we can’t measure over all inputs, we do our best and measure the empirical risk over the data we have. Optimization How to minimize loss? Descend the gradient. Stochastic Gradient Descent (SGD) Gradient Descent Why? - Too much data makes the sum expensive - Unreasonable Stochastic Gradient Descent effectiveness of randomization Why Deep Learning? The Unreasonable Effectiveness of Deep Features Classes separate in the deep representations and transfer to many tasks. [DeCAF] [Zeiler-Fergus] Why Deep Learning? The Unreasonable Effectiveness of Deep Features Maximal activations of pool5 units [R-CNN] conv5 DeConv visualization Rich visual structure of features deep in hierarchy. [Zeiler-Fergus] Why Deep Learning? The Unreasonable Effectiveness of Deep Features 1st layer filters image patches that strongly activate 1st layer filters [Zeiler-Fergus] Each 3x3 cell shows the top 9 image patches that activate a given feature in this layer. Note the increase in visual complexity of feature activations as we go from “shallow” to “deep” layers. [Zeiler-Fergus] [Zeiler-Fergus] [Zeiler-Fergus] [Zeiler-Fergus] Why Deep Learning? The Unreasonable Effectiveness of Deep Features Applications! - vision - speech - text - ..
    [Show full text]
  • Terrain-Adaptive Locomotion Skills Using Deep Reinforcement Learning
    Terrain-Adaptive Locomotion Skills Using Deep Reinforcement Learning Xue Bin Peng, Glen Berseth, Michiel van de Panne⇤ University of British Columbia Figure 1: Terrain traversal using a learned actor-critic ensemble. The color-coding of the center-of-mass trajectory indicates the choice of actor used for each leap. Abstract In practice, a number of challenges need to be overcome when ap- plying the RL framework to problems with continuous and high- Reinforcement learning offers a promising methodology for devel- dimensional states and actions, as required by movement skills. A oping skills for simulated characters, but typically requires working control policy needs to select the best actions for the distribution with sparse hand-crafted features. Building on recent progress in of states that will be encountered, but this distribution is often not deep reinforcement learning (DeepRL), we introduce a mixture of known in advance. Similarly, the distribution of actions that will actor-critic experts (MACE) approach that learns terrain-adaptive prove to be useful for these states is also seldom known in advance. dynamic locomotion skills using high-dimensional state and terrain Furthermore, the state-and-action distributions are not static in na- descriptions as input, and parameterized leaps or steps as output ture; as changes are made to the control policy, new states may actions. MACE learns more quickly than a single actor-critic ap- be visited, and, conversely, the best possible policy may change as proach and results in actor-critic experts that exhibit specialization. new actions are introduced. It is furthermore not obvious how to Additional elements of our solution that contribute towards efficient best represent the state of a character and its environment.
    [Show full text]