DOCSLIB.ORG

Representation Learning for Control

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Representation Learning for Control Representation Learning for Control A Thesis Presented by Kevin James Doty to The Department of Electrical and Computer Engineering in partial fulfillment of the requirements for the degree of Master of Science in Electrical and Computer Engineering Northeastern University Boston, Massachusetts May 2019 To my family. i Contents List of Figures iv List of Tables v List of Acronyms vi Acknowledgments vii Abstract of the Thesis viii 1 Introduction 1 2 Background 3 2.1 Markov Decision Process . 3 2.2 Solving MDP’s . 4 2.2.1 Q-Learning . 5 2.3 Convolutional Neural Network . 6 2.4 Deep Q-Learning . 7 2.4.1 Experience Replay . 7 2.4.2 Policy and Target Network . 8 2.5 Autoencoder . 8 2.6 Variational Autoencoder . 9 3 Methodology 11 3.1 Setup . 11 3.2 Puck-Stack Environment . 11 3.3 Data Generation . 12 3.4 Proposed Architectures . 12 3.4.1 Reconstruction Architectures . 13 3.4.2 Predictive Variational Autoencoder . 15 3.4.3 Inverse Models . 16 3.5 Training Procedure . 18 ii 3.6 Evaluation . 18 4 Results 19 5 Conclusions 23 5.1 The Motivation for Representation Learning . 23 5.2 Future Work . 23 5.2.1 Partial Observability . 24 5.2.2 Representation Conducive to Planning . 24 Bibliography 25 A VAE Loss Derivation 27 B Architecture Building Block Summary 29 iii List of Figures 2.1 MDP Agent Environment Interaction . 3 3.1 Successful Sequence in PuckStack . 12 3.2 VAE Structure . 15 3.3 Predictive VAE Structure . 16 3.4 Inverse Model Structure . 17 3.5 Variational Inverse Model Structure . 18 4.1 Comparing latent controller performances to the fully convolutional DQN and the VAE latent-DQN, min-max-mean training curves over five training trials . 20 iv List of Tables 4.1 Optimized Representation Learning Architecture Parameters . 22 v List of Acronyms AE Autoencoder. A neural network structure that maps an input to itself through a bottle- neck layer. CNN Convolutional Neural Network. A network that filters an input feature map by applying the correlation operation between kernels with learned weights and the input features. DQN Deep Q-Network. An extension of traditional Q-learning to continuous state spaces by using neural networks as Q-function approximators. MDP Markov Decision Process. A discrete time control process with perfect information. At each time step a decision making agent perceives a state, selects an action, and then progresses to a new state, optionally receiving a reward. PCA Principal Component Analysis. A statistical procedure for converting data into a set of uncorrelated features. POMDP Partially Observable Markov Decision Process. A special case of the markov decision process, in which the agent receives observations that are not the true state, and do not contain perfect information. SARS (State, Action, Reward, Next State) VAE Variational Autoencoder. A probabilistic autoencoding structure in which the encoder is trained to approximate the posterior distribution of the data generation process. vi Acknowledgments I would like to thank my family and friends for their support and my advisor Rob Platt for the insightful discussions and guidance. I would also like to express my appreciation for the robotics community of NEU at large. The past couple of years have truly been fulfilling. It has been my pleasure to be here learning with all of you. vii Abstract of the Thesis Representation Learning for Control by Kevin James Doty Master of Science in Electrical and Computer Engineering Northeastern University, May 2019 Dr. Robert Platt, Advisor State representation learning finds an embedding from a high dimensional observation space to a lower dimensional and information dense state space, without supervision. Effective state representation learning can improve the sample efficiency and stability of downstream tasks such as reinforcement learning for control. In this work, two autoencoder variants and two inverse models are explored as representation learning architectures. The DQN algorithm is used to learn control in the resulting latent spaces. Latent-control performance is benchmarked against a standard fully convolutional DQN on a downstream control task. viii Chapter 1 Introduction One of the fundamental challenges of reinforcement learning is learning control policies in high dimensional observation spaces. This is particularly important in robotics applica- tions, where we would like to develop behaviors that are responsive to raw sensory data such as images, point clouds or audio. Often these raw data contain substantial noise and features irrelevant to the task. Learning in these conditions can require excessive amounts of data and time, and is prone to instability. One idea that can be applied to get around the curse of dimensionality in raw data is state representation learning. In state representation learning the structure in information, that is gathered passively or actively, is exploited to learn mappings from a high dimensional observation space to a compact and meaningful state space. The learned mappings can then be used to pre-process raw data before sending it to a reinforcement learning algorithm as state. This process can greatly simplify the problem of policy learning. In this work, two forms of representation learning are explored, the autoencoder and the inverse model. The autoencoder has a fully unsupervised training process in which an input is mapped to itself through an information bottleneck. The structure is trained to reconstruct its input after compressing the information down to a predefined dimensionality. The inverse model is a self-supervised technique. Here, agent actions and observations influenced by those actions are recorded simultaneously. A siamese network structure is then given sequential observations as intput and trained to predict the action that occurred between them. This forces an encoder to learn features relevant for discerning the effect of 1 CHAPTER 1. INTRODUCTION actions in the environment. Autoencoders and inverse models have been applied in various settings. This work seeks to understand the relative performance of the two architectures, as well as to test two novel architectures based on the autoencoding and inverse model paradigms. To accomplish this, these architectures are trained on data collected in the PuckStack environment. Deep Q-Network (DQN) agents are then trained in the latent spaces of the resulting encoders. The number of training episodes until convergence, training stability, and performance of these controllers are compared to each other and to a DQN baseline agent that learns directly from environment pixels. The goals of this work are to produce further understanding of the abilities and relative performances of the tested representation learning techniques, and to inform the design of these architectures for future applications. Effective state representation learning has the potential to solve many of the difficult aspects of control learning from raw data, in an unsupervised or self-supervised manner. The long term goals for this field are to reduce the amount of data and time necessary to learn control policies, and to increase the stability and overall capability of these policies. With some progress, it may become feasible to train policies on physical robots in the lab in days or hours. 2 Chapter 2 Background 2.1 Markov Decision Process The Markov Decision Process (MDP) is a mathematical framework for modeling se- quential decision making. The MDP involves the notions of an agent and an environment that interact through action selection, state transitions, and reward signals. The decision making agent perceives the state of the environment, this agent then chooses an action, progresses to a new state and receives a reward signal. Figure 2.1: MDP Agent Environment Interaction (Source [1]) An MDP can be formally defined by a state space, action space, a state transition function, and a reward function. The state transition function returns a probability distribution over next state given a current state and action. The reward function returns a distribution over rewards given a state and action. 3 CHAPTER 2. BACKGROUND In this work we consider only finite episodic MDP’s in which the state space and action space are finite, and the agents interactions with the environment consist of a series of finite length sequences. The interaction of the agent and the environment continues until the agent encounters some terminal state. A traversal through the environment from initial state to terminal state is referred to as an episode. 2.2 Solving MDP’s The goal of the agent is to maximize the cumulative reward it receives from the envi- ronment during an episode. This can be formalized as the return defined in 2.1, where γ is a discount factor with a value in the interval [0, 1). This factor prevents the sum from equaling infinity, and can be used to weight the importance of short term verse long term rewards in a particular application. This formula can be regrouped as in 2.2 so it can then be expressed recursively 2.3. 2 3 Gt , Rt+1 + γRt+2 + γ Rt+3 + γ Rt+4 + ··· (2.1) 2 = Rt+1 + γ(Rt+2 + γRt+3 + γ Rt+4 + ··· ) (2.2) = Rt+1 + γGt+1 (2.3) Action selection in an MDP dictates not only reward received at the current time step, but also subsequent states, and therefore rewards that can be received in future time steps. In solving an MDP, current and future rewards must be considered. This can involve selecting low reward actions in the present to gain access to higher rewards in future steps. The function the agent uses to select actions given states is referred to as a policy. This policy can be probabilistic or deterministic. We want to learn an optimal policy, or a policy that achieves the maximum return possible in the environment. There are many different ways of solving for optimal policies in the reinforcement learning literature.
Load more

Recommended publications
  • Disentangled Variational Auto-Encoder for Semi-Supervised Learning
    Information Sciences 482 (2019) 73–85 Contents lists available at ScienceDirect Information Sciences journal homepage: www.elsevier.com/locate/ins Disentangled Variational Auto-Encoder for semi-supervised learning ∗ Yang Li a, Quan Pan a, Suhang Wang c, Haiyun Peng b, Tao Yang a, Erik Cambria b, a School of Automation, Northwestern Polytechnical University, China b School of Computer Science and Engineering, Nanyang Technological University, Singapore c College of Information Sciences and Technology, Pennsylvania State University, USA a r t i c l e i n f o a b s t r a c t Article history: Semi-supervised learning is attracting increasing attention due to the fact that datasets Received 5 February 2018 of many domains lack enough labeled data. Variational Auto-Encoder (VAE), in particu- Revised 23 December 2018 lar, has demonstrated the benefits of semi-supervised learning. The majority of existing Accepted 24 December 2018 semi-supervised VAEs utilize a classifier to exploit label information, where the param- Available online 3 January 2019 eters of the classifier are introduced to the VAE. Given the limited labeled data, learn- Keywords: ing the parameters for the classifiers may not be an optimal solution for exploiting label Semi-supervised learning information. Therefore, in this paper, we develop a novel approach for semi-supervised Variational Auto-encoder VAE without classifier. Specifically, we propose a new model called Semi-supervised Dis- Disentangled representation entangled VAE (SDVAE), which encodes the input data into disentangled representation Neural networks and non-interpretable representation, then the category information is directly utilized to regularize the disentangled representation via the equality constraint.
    [Show full text]
  • Training Autoencoders by Alternating Minimization
    Under review as a conference paper at ICLR 2018 TRAINING AUTOENCODERS BY ALTERNATING MINI- MIZATION Anonymous authors Paper under double-blind review ABSTRACT We present DANTE, a novel method for training neural networks, in particular autoencoders, using the alternating minimization principle. DANTE provides a distinct perspective in lieu of traditional gradient-based backpropagation techniques commonly used to train deep networks. It utilizes an adaptation of quasi-convex optimization techniques to cast autoencoder training as a bi-quasi-convex optimiza- tion problem. We show that for autoencoder configurations with both differentiable (e.g. sigmoid) and non-differentiable (e.g. ReLU) activation functions, we can perform the alternations very effectively. DANTE effortlessly extends to networks with multiple hidden layers and varying network configurations. In experiments on standard datasets, autoencoders trained using the proposed method were found to be very promising and competitive to traditional backpropagation techniques, both in terms of quality of solution, as well as training speed. 1 INTRODUCTION For much of the recent march of deep learning, gradient-based backpropagation methods, e.g. Stochastic Gradient Descent (SGD) and its variants, have been the mainstay of practitioners. The use of these methods, especially on vast amounts of data, has led to unprecedented progress in several areas of artificial intelligence. On one hand, the intense focus on these techniques has led to an intimate understanding of hardware requirements and code optimizations needed to execute these routines on large datasets in a scalable manner. Today, myriad off-the-shelf and highly optimized packages exist that can churn reasonably large datasets on GPU architectures with relatively mild human involvement and little bootstrap effort.
    [Show full text]
  • A Deep Hierarchical Variational Autoencoder
    NVAE: A Deep Hierarchical Variational Autoencoder Arash Vahdat, Jan Kautz NVIDIA {avahdat, jkautz}@nvidia.com Abstract Normalizing flows, autoregressive models, variational autoencoders (VAEs), and deep energy-based models are among competing likelihood-based frameworks for deep generative learning. Among them, VAEs have the advantage of fast and tractable sampling and easy-to-access encoding networks. However, they are cur- rently outperformed by other models such as normalizing flows and autoregressive models. While the majority of the research in VAEs is focused on the statistical challenges, we explore the orthogonal direction of carefully designing neural archi- tectures for hierarchical VAEs. We propose Nouveau VAE (NVAE), a deep hierar- chical VAE built for image generation using depth-wise separable convolutions and batch normalization. NVAE is equipped with a residual parameterization of Normal distributions and its training is stabilized by spectral regularization. We show that NVAE achieves state-of-the-art results among non-autoregressive likelihood-based models on the MNIST, CIFAR-10, CelebA 64, and CelebA HQ datasets and it provides a strong baseline on FFHQ. For example, on CIFAR-10, NVAE pushes the state-of-the-art from 2.98 to 2.91 bits per dimension, and it produces high-quality images on CelebA HQ as shown in Fig. 1. To the best of our knowledge, NVAE is the first successful VAE applied to natural images as large as 256×256 pixels. The source code is available at https://github.com/NVlabs/NVAE. 1 Introduction The majority of the research efforts on improving VAEs [1, 2] is dedicated to the statistical challenges, such as reducing the gap between approximate and true posterior distributions [3, 4, 5, 6, 7, 8, 9, 10], formulating tighter bounds [11, 12, 13, 14], reducing the gradient noise [15, 16], extending VAEs to discrete variables [17, 18, 19, 20, 21, 22, 23], or tackling posterior collapse [24, 25, 26, 27].
    [Show full text]
  • Explainable Deep Learning Models in Medical Image Analysis
    Journal of Imaging Review Explainable Deep Learning Models in Medical Image Analysis Amitojdeep Singh 1,2,* , Sourya Sengupta 1,2 and Vasudevan Lakshminarayanan 1,2 1 Theoretical and Experimental Epistemology Laboratory, School of Optometry and Vision Science, University of Waterloo, Waterloo, ON N2L 3G1, Canada; [email protected] (S.S.); [email protected] (V.L.) 2 Department of Systems Design Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada * Correspondence: [email protected] Received: 28 May 2020; Accepted: 17 June 2020; Published: 20 June 2020 Abstract: Deep learning methods have been very effective for a variety of medical diagnostic tasks and have even outperformed human experts on some of those. However, the black-box nature of the algorithms has restricted their clinical use. Recent explainability studies aim to show the features that influence the decision of a model the most. The majority of literature reviews of this area have focused on taxonomy, ethics, and the need for explanations. A review of the current applications of explainable deep learning for different medical imaging tasks is presented here. The various approaches, challenges for clinical deployment, and the areas requiring further research are discussed here from a practical standpoint of a deep learning researcher designing a system for the clinical end-users. Keywords: explainability; explainable AI; XAI; deep learning; medical imaging; diagnosis 1. Introduction Computer-aided diagnostics (CAD) using artificial intelligence (AI) provides a promising way to make the diagnosis process more efficient and available to the masses. Deep learning is the leading artificial intelligence (AI) method for a wide range of tasks including medical imaging problems.
    [Show full text]
  • Turbo Autoencoder: Deep Learning Based Channel Codes for Point-To-Point Communication Channels
    Turbo Autoencoder: Deep learning based channel codes for point-to-point communication channels Hyeji Kim Himanshu Asnani Yihan Jiang Samsung AI Center School of Technology ECE Department Cambridge and Computer Science University of Washington Cambridge, United Tata Institute of Seattle, United States Kingdom Fundamental Research [email protected] [email protected] Mumbai, India [email protected] Sewoong Oh Pramod Viswanath Sreeram Kannan Allen School of ECE Department ECE Department Computer Science & University of Illinois at University of Washington Engineering Urbana Champaign Seattle, United States University of Washington Illinois, United States [email protected] Seattle, United States [email protected] [email protected] Abstract Designing codes that combat the noise in a communication medium has remained a significant area of research in information theory as well as wireless communica- tions. Asymptotically optimal channel codes have been developed by mathemati- cians for communicating under canonical models after over 60 years of research. On the other hand, in many non-canonical channel settings, optimal codes do not exist and the codes designed for canonical models are adapted via heuristics to these channels and are thus not guaranteed to be optimal. In this work, we make significant progress on this problem by designing a fully end-to-end jointly trained neural encoder and decoder, namely, Turbo Autoencoder (TurboAE), with the following contributions: (a) under moderate block lengths, TurboAE approaches state-of-the-art performance under canonical channels; (b) moreover, TurboAE outperforms the state-of-the-art codes under non-canonical settings in terms of reliability. TurboAE shows that the development of channel coding design can be automated via deep learning, with near-optimal performance.
    [Show full text]
  • Double Backpropagation for Training Autoencoders Against Adversarial Attack
    1 Double Backpropagation for Training Autoencoders against Adversarial Attack Chengjin Sun, Sizhe Chen, and Xiaolin Huang, Senior Member, IEEE Abstract—Deep learning, as widely known, is vulnerable to adversarial samples. This paper focuses on the adversarial attack on autoencoders. Safety of the autoencoders (AEs) is important because they are widely used as a compression scheme for data storage and transmission, however, the current autoencoders are easily attacked, i.e., one can slightly modify an input but has totally different codes. The vulnerability is rooted the sensitivity of the autoencoders and to enhance the robustness, we propose to adopt double backpropagation (DBP) to secure autoencoder such as VAE and DRAW. We restrict the gradient from the reconstruction image to the original one so that the autoencoder is not sensitive to trivial perturbation produced by the adversarial attack. After smoothing the gradient by DBP, we further smooth the label by Gaussian Mixture Model (GMM), aiming for accurate and robust classification. We demonstrate in MNIST, CelebA, SVHN that our method leads to a robust autoencoder resistant to attack and a robust classifier able for image transition and immune to adversarial attack if combined with GMM. Index Terms—double backpropagation, autoencoder, network robustness, GMM. F 1 INTRODUCTION N the past few years, deep neural networks have been feature [9], [10], [11], [12], [13], or network structure [3], [14], I greatly developed and successfully used in a vast of fields, [15]. such as pattern recognition, intelligent robots, automatic Adversarial attack and its defense are revolving around a control, medicine [1]. Despite the great success, researchers small ∆x and a big resulting difference between f(x + ∆x) have found the vulnerability of deep neural networks to and f(x).
    [Show full text]
  • Generative Models
    Lecture 11: Generative Models Fei-Fei Li & Justin Johnson & Serena Yeung Lecture 11 - 1 May 9, 2019 Administrative ● A3 is out. Due May 22. ● Milestone is due next Wednesday. ○ Read Piazza post for milestone requirements. ○ Need to Finish data preprocessing and initial results by then. ● Don't discuss exam yet since people are still taking it. Fei-Fei Li & Justin Johnson & Serena Yeung Lecture 11 -2 May 9, 2019 Overview ● Unsupervised Learning ● Generative Models ○ PixelRNN and PixelCNN ○ Variational Autoencoders (VAE) ○ Generative Adversarial Networks (GAN) Fei-Fei Li & Justin Johnson & Serena Yeung Lecture 11 - 3 May 9, 2019 Supervised vs Unsupervised Learning Supervised Learning Data: (x, y) x is data, y is label Goal: Learn a function to map x -> y Examples: Classification, regression, object detection, semantic segmentation, image captioning, etc. Fei-Fei Li & Justin Johnson & Serena Yeung Lecture 11 - 4 May 9, 2019 Supervised vs Unsupervised Learning Supervised Learning Data: (x, y) x is data, y is label Cat Goal: Learn a function to map x -> y Examples: Classification, regression, object detection, Classification semantic segmentation, image captioning, etc. This image is CC0 public domain Fei-Fei Li & Justin Johnson & Serena Yeung Lecture 11 - 5 May 9, 2019 Supervised vs Unsupervised Learning Supervised Learning Data: (x, y) x is data, y is label Goal: Learn a function to map x -> y Examples: Classification, DOG, DOG, CAT regression, object detection, semantic segmentation, image Object Detection captioning, etc. This image is CC0 public domain Fei-Fei Li & Justin Johnson & Serena Yeung Lecture 11 - 6 May 9, 2019 Supervised vs Unsupervised Learning Supervised Learning Data: (x, y) x is data, y is label Goal: Learn a function to map x -> y Examples: Classification, GRASS, CAT, TREE, SKY regression, object detection, semantic segmentation, image Semantic Segmentation captioning, etc.
    [Show full text]
  • Artificial Intelligence Applied to Electromechanical Monitoring, A
    ARTIFICIAL INTELLIGENCE APPLIED TO ELECTROMECHANICAL MONITORING, A PERFORMANCE ANALYSIS Erasmus project Authors: Staš Osterc Mentor: Dr. Miguel Delgado Prieto, Dr. Francisco Arellano Espitia 1/5/2020 P a g e II ANNEX VI – DECLARACIÓ D’HONOR P a g e II I declare that, the work in this Master Thesis / Degree Thesis (choose one) is completely my own work, no part of this Master Thesis / Degree Thesis (choose one) is taken from other people’s work without giving them credit, all references have been clearly cited, I’m authorised to make use of the company’s / research group (choose one) related information I’m providing in this document (select when it applies). I understand that an infringement of this declaration leaves me subject to the foreseen disciplinary actions by The Universitat Politècnica de Catalunya - BarcelonaTECH. ___________________ __________________ ___________ Student Name Signature Date Title of the Thesis : _________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ P a g e II Contents Introduction........................................................................................................................................5 Abstract ..........................................................................................................................................5 Aim .................................................................................................................................................6
    [Show full text]
  • Infinite Variational Autoencoder for Semi-Supervised Learning
    Infinite Variational Autoencoder for Semi-Supervised Learning M. Ehsan Abbasnejad Anthony Dick Anton van den Hengel The University of Adelaide {ehsan.abbasnejad, anthony.dick, anton.vandenhengel}@adelaide.edu.au Abstract with whatever labelled data is available to train a discrimi- native model for classification. This paper presents an infinite variational autoencoder We demonstrate that our infinite VAE outperforms both (VAE) whose capacity adapts to suit the input data. This the classical VAE and standard classification methods, par- is achieved using a mixture model where the mixing coef- ticularly when the number of available labelled samples is ficients are modeled by a Dirichlet process, allowing us to small. This is because the infinite VAE is able to more ac- integrate over the coefficients when performing inference. curately capture the distribution of the unlabelled data. It Critically, this then allows us to automatically vary the therefore provides a generative model that allows the dis- number of autoencoders in the mixture based on the data. criminative model, which is trained based on its output, to Experiments show the flexibility of our method, particularly be more effectively learnt using a small number of samples. for semi-supervised learning, where only a small number of The main contribution of this paper is twofold: (1) we training samples are available. provide a Bayesian non-parametric model for combining autoencoders, in particular variational autoencoders. This bridges the gap between non-parametric Bayesian meth- 1. Introduction ods and the deep neural networks; (2) we provide a semi- supervised learning approach that utilizes the infinite mix- The Variational Autoencoder (VAE) [18] is a newly in- ture of autoencoders learned by our model for prediction troduced tool for unsupervised learning of a distribution x x with from a small number of labeled examples.
    [Show full text]
  • Unsupervised Speech Representation Learning Using Wavenet Autoencoders Jan Chorowski, Ron J
    1 Unsupervised speech representation learning using WaveNet autoencoders Jan Chorowski, Ron J. Weiss, Samy Bengio, Aaron¨ van den Oord Abstract—We consider the task of unsupervised extraction speaker gender and identity, from phonetic content, properties of meaningful latent representations of speech by applying which are consistent with internal representations learned autoencoding neural networks to speech waveforms. The goal by speech recognizers [13], [14]. Such representations are is to learn a representation able to capture high level semantic content from the signal, e.g. phoneme identities, while being desired in several tasks, such as low resource automatic speech invariant to confounding low level details in the signal such as recognition (ASR), where only a small amount of labeled the underlying pitch contour or background noise. Since the training data is available. In such scenario, limited amounts learned representation is tuned to contain only phonetic content, of data may be sufficient to learn an acoustic model on the we resort to using a high capacity WaveNet decoder to infer representation discovered without supervision, but insufficient information discarded by the encoder from previous samples. Moreover, the behavior of autoencoder models depends on the to learn the acoustic model and a data representation in a fully kind of constraint that is applied to the latent representation. supervised manner [15], [16]. We compare three variants: a simple dimensionality reduction We focus on representations learned with autoencoders bottleneck, a Gaussian Variational Autoencoder (VAE), and a applied to raw waveforms and spectrogram features and discrete Vector Quantized VAE (VQ-VAE). We analyze the quality investigate the quality of learned representations on LibriSpeech of learned representations in terms of speaker independence, the ability to predict phonetic content, and the ability to accurately re- [17].
    [Show full text]
  • Variational Autoencoder for End-To-End Control of Autonomous Driving with Novelty Detection and Training De-Biasing
    Variational Autoencoder for End-to-End Control of Autonomous Driving with Novelty Detection and Training De-biasing Alexander Amini1, Wilko Schwarting1, Guy Rosman2, Brandon Araki1, Sertac Karaman3, Daniela Rus1 Abstract— This paper introduces a new method for end-to- Uncertainty Estimation end training of deep neural networks (DNNs) and evaluates & Novelty Detection it in the context of autonomous driving. DNN training has Uncertainty been shown to result in high accuracy for perception to action Propagate learning given sufficient training data. However, the trained models may fail without warning in situations with insufficient or biased training data. In this paper, we propose and evaluate Encoder a novel architecture for self-supervised learning of latent variables to detect the insufficiently trained situations. Our method also addresses training data imbalance, by learning a set of underlying latent variables that characterize the training Dataset Debiasing data and evaluate potential biases. We show how these latent Weather Adjacent Road Steering distributions can be leveraged to adapt and accelerate the (Snow) Vehicles Surface Control training pipeline by training on only a fraction of the total Command Resampled data dataset. We evaluate our approach on a challenging dataset distribution for driving. The data is collected from a full-scale autonomous Unsupervised Latent vehicle. Our method provides qualitative explanation for the Variables Sample Efficient latent variables learned in the model. Finally, we show how Accelerated Training our model can be additionally trained as an end-to-end con- troller, directly outputting a steering control command for an Fig. 1: Semi-supervised end-to-end control. An encoder autonomous vehicle.
    [Show full text]
  • An Introduction to Variational Autoencoders Full Text Available At
    Full text available at: http://dx.doi.org/10.1561/2200000056 An Introduction to Variational Autoencoders Full text available at: http://dx.doi.org/10.1561/2200000056 Other titles in Foundations and Trends R in Machine Learning Computational Optimal Transport Gabriel Peyre and Marco Cuturi ISBN: 978-1-68083-550-2 An Introduction to Deep Reinforcement Learning Vincent Francois-Lavet, Peter Henderson, Riashat Islam, Marc G. Bellemare and Joelle Pineau ISBN: 978-1-68083-538-0 An Introduction to Wishart Matrix Moments Adrian N. Bishop, Pierre Del Moral and Angele Niclas ISBN: 978-1-68083-506-9 A Tutorial on Thompson Sampling Daniel J. Russo, Benjamin Van Roy, Abbas Kazerouni, Ian Osband and Zheng Wen ISBN: 978-1-68083-470-3 Full text available at: http://dx.doi.org/10.1561/2200000056 An Introduction to Variational Autoencoders Diederik P. Kingma Google [email protected] Max Welling University of Amsterdam Qualcomm [email protected] Boston — Delft Full text available at: http://dx.doi.org/10.1561/2200000056 Foundations and Trends R in Machine Learning Published, sold and distributed by: now Publishers Inc. PO Box 1024 Hanover, MA 02339 United States Tel. +1-781-985-4510 www.nowpublishers.com [email protected] Outside North America: now Publishers Inc. PO Box 179 2600 AD Delft The Netherlands Tel. +31-6-51115274 The preferred citation for this publication is D. P. Kingma and M. Welling. An Introduction to Variational Autoencoders. Foundations and Trends R in Machine Learning, vol. 12, no. 4, pp. 307–392, 2019. ISBN: 978-1-68083-623-3 c 2019 D.
    [Show full text]