Data augmentation and image understanding

Abstract

Interdisciplinary research is often at the core of scientific progress. This dissertation explores some advantageous synergies between machine learning, cognitive science and neuroscience. In particular, this thesis focuses on vision and images. The human visual system has been widely studied from both behavioural and neuroscientific points of view, as vision is the dominant sense of most people. In turn, machine vision has also been an active area of research, currently dominated by the use of artificial neural networks. This work focuses on learning representations that are more aligned with visual perception and the biological vision. For that purpose, I have studied tools and aspects from cognitive science and computational neuroscience, and attempted to incorporate them into machine learning models of vision. A central subject of this dissertation is data augmentation, a commonly used technique for training artificial neural networks to augment the size of data sets through transformations of the images. Although often overlooked, data augmentation implements transformations that are perceptually plausible, since they correspond to the transformations we see in our visual world -- changes in viewpoint or illumination, for instance. Furthermore, neuroscientists have found that the brain invariantly represents objects under these transformations. Throughout this dissertation, I use these insights to analyse data augmentation as a particularly useful inductive bias, a more effective regularisation method for artificial neural networks, and as the framework to analyse and improve the invariance of vision models to perceptually plausible transformations. Overall, this work aims to shed more light on the properties of data augmentation and demonstrate the potential of interdisciplinary research.

arXiv commentDigital version of the PhD thesis by Alex Hernandez-Garcia, defended on the 27th of November of 2020 at the Institute of Cognitive Science of the University of Osnabrueck, Germany. PhD advisor: Prof. Peter Koenig. Contributors are acknowledged at the beginning of each chapter

References in corpus (30)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• mixup: Beyond Empirical Risk Minimization
• Improved Regularization of Convolutional Neural Networks with Cutout
• The Effectiveness of Data Augmentation in Image Classification using Deep Learning
• Shortcut Learning in Deep Neural Networks
• Reconciling modern machine learning practice and the bias-variance trade-off
• Deep Neural Networks Reveal a Gradient in the Complexity of Neural Representations across the Brain's Ventral Visual Pathway
• Data Augmentation Generative Adversarial Networks
• Opening the Black Box of Deep Neural Networks via Information
• AutoAugment: Learning Augmentation Policies from Data
• Group Equivariant Convolutional Networks
• Learning Representations and Generative Models for 3D Point Clouds
• Smart Augmentation - Learning an Optimal Data Augmentation Strategy
• Adversarial Examples Are Not Bugs, They Are Features
• Making Convolutional Networks Shift-Invariant Again
• Towards Biologically Plausible Deep Learning
• DeepGaze II: Reading fixations from deep features trained on object recognition
• Dropout Training as Adaptive Regularization
• Measuring the tendency of CNNs to Learn Surface Statistical Regularities
• Quantifying the Carbon Emissions of Machine Learning
• Data augmentation instead of explicit regularization
• Dropout as data augmentation
• Dataset Augmentation in Feature Space
• Augment your batch: better training with larger batches
• High Frequency Component Helps Explain the Generalization of Convolutional Neural Networks
• On the Benefits of Invariance in Neural Networks
• Adversarial Training Versus Weight Decay
• Neural Networks Regularization Through Class-wise Invariant Representation Learning
• Does Data Augmentation Lead to Positive Margin?