We study Label-Smoothing as a means for improving adversarial robustness of supervised deep-learning models. After establishing a thorough and unified framework, we propose several novel Label-Smoothing methods: adversarial, Boltzmann and second-best Label-Smoothing methods. On various datasets (MNIST, CI-FAR10, SVHN) and models (linear models, MLPs, LeNet, ResNet), we show that these methods improve adversarial robustness against a variety of attacks (FGSM, BIM, DeepFool, Carlini-Wargner) by better taking account of the dataset geometry. These proposed Label-Smoothing methods have two main advantages: they can be implemented as a modified cross-entropy loss, thus do not require any modifications of the network architecture nor do they lead to increased training times, and they improve both standard and adversarial accuracy.

This article presents a continuous model for hierarchical networks based on a combination of mathematically derived models of receptive fields and biologically inspired computations. Based on a functional model of complex cells in terms of an oriented quasi quadrature combination of first- and second-order directional Gaussian derivatives, we couple such primitive computations in cascade over combinatorial expansions over image orientations. Scale-space properties of the computational primitives are analysed and it is shown that the resulting representation allows for provable scale and rotation covariance. A prototype application to texture analysis is developed and it is demonstrated that a simplified mean-reduced representation of the resulting QuasiQuadNet leads to promising experimental results on three texture datasets.

Model selection is treated as a standard performance boosting step in many machine learning applications. Once all other properties of a learning problem are fixed, the model is selected by grid search on a held-out validation set. This is strictly inapplicable to active learning. Within the standardized workflow, the acquisition function is chosen among available heuristics a priori, and its success is observed only after the labeling budget is already exhausted. More importantly, none of the earlier studies report a unique consistently successful acquisition heuristic to the extent to stand out as the unique best choice. We present a method to break this vicious circle by defining the acquisition function as a learning predictor and training it by reinforcement feedback collected from each labeling round. As active learning is a scarce data regime, we bootstrap from a well-known heuristic that filters the bulk of data points on which all heuristics would agree, and learn a policy to warp the top portion of this ranking in the most beneficial way for the character of a specific data distribution. Our system consists of a Bayesian neural net, the predictor, a bootstrap acquisition function, a probabilistic state definition, and another Bayesian policy network that can effectively incorporate this input distribution. We observe on three benchmark data sets that our method always manages to either invent a new superior acquisition function or to adapt itself to the a priori unknown best performing heuristic for each specific data set.

Graph representation on large-scale bipartite graphs is central for a variety of applications, ranging from social network analysis to recommendation system development. Existing methods exhibit two key drawbacks: 1. unable to characterize the inconsistency of the node features within the bipartite-specific structure; 2. unfriendly to support large-scale bipartite graphs. To this end, we propose ABCGraph, a scalable model for unsupervised learning on large-scale bipartite graphs. At its heart, ABCGraph utilizes the proposed Bipartite Graph Convolutional Network (BGCN) as the encoder and adversarial learning as the training loss to learn representations from nodes in two different domains and bipartite structures, in an unsupervised manner. Moreover, we devise a cascaded architecture to capture the multi-hop relationship in bipartite structure and improves the scalability as well. Extensive experiments on multiple datasets of varying scales verify the effectiveness of ABCGraph compared to state-of-the-arts. For the experiment on a real-world large-scale bipartite graph system, fast training speed and low memory cost demonstrate the scalability of ABCGraph model.

When translating natural language questions into SQL queries to answer questions from a database, we would like our methods to generalize to domains and database schemas outside of the training set. To handle complex questions and database schemas with a neural encoder-decoder paradigm, it is critical to properly encode the schema as part of the input with the question. In this paper, we use relation-aware self-attention within the encoder so that it can reason about how the tables and columns in the provided schema relate to each other and use this information in interpreting the question. We achieve significant gains on the recently-released Spider dataset with 42.94% exact match accuracy, compared to the 18.96% reported in published work.

Graph neural networks have become increasingly popular in recent years due to their ability to naturally encode relational input data and their ability to scale to large graphs by operating on a sparse representation of graph adjacency matrices. As we look to scale up these models using custom hardware, a natural assumption would be that we need hardware tailored to sparse operations and/or dynamic control flow. In this work, we question this assumption by scaling up sparse graph neural networks using a platform targeted at dense computation on fixed-size data. Drawing inspiration from optimization of numerical algorithms on sparse matrices, we develop techniques that enable training the sparse graph neural network model from Allamanis et al. [2018] in 13 minutes using a 512-core TPUv2 Pod, whereas the original training takes almost a day.

Training generative models like generative adversarial networks (GANs) and normalizing flows is challenging for noisy data. A novel curriculum learning algorithm pertaining to clustering is proposed to address this issue in this paper. The curriculum construction is based on the centrality of underlying clusters in data points. The data points of high centrality takes priority of being fed into generative models during training. To make our algorithm scalable to large-scale data, the active set is devised, in the sense that every round of training proceeds only on an active subset containing a small fraction of already trained data and the incremental data of lower centrality. Moreover, the geometric analysis is presented to interpret the necessity of cluster curriculum for generative models. The experiments on cat and human-face data validate that our algorithm is able to learn the optimal generative models (e.g. ProGAN and Glow) with respect to specified quality metrics for noisy data. An interesting finding is that the optimal cluster curriculum is closely related to the critical point of a geometric percolation process formulated in the paper.

Hyperparameter tuning is an omnipresent problem in machine learning as it is an integral aspect of obtaining the state-of-the-art performance for any model. Most often, hyperparameters are optimized just by training a model on a grid of possible hyperparameter values and taking the one that performs best on a validation sample (grid search). More recently, methods have been introduced that build a so-called surrogate model that predicts the validation loss for a specific hyperparameter setting, model and dataset and then sequentially select the next hyperparameter to test, based on a heuristic function of the expected value and the uncertainty of the surrogate model called acquisition function (sequential model-based Bayesian optimization, SMBO). In this paper we model the hyperparameter optimization problem as a sequential decision problem, which hyperparameter to test next, and address it with reinforcement learning. This way our model does not have to rely on a heuristic acquisition function like SMBO, but can learn which hyperparameters to test next based on the subsequent reduction in validation loss they will eventually lead to, either because they yield good models themselves or because they allow the hyperparameter selection policy to build a better surrogate model that is able to choose better hyperparameters later on. Experiments on a large battery of 50 data sets demonstrate that our method outperforms the state-of-the-art approaches for hyperparameter learning.

In this paper, we performed two types of software experiments to study the numerosity classification (subitizing) in humans and machines. Experiments focus on a particular kind of task is referred to as Semantic MNIST or simply SMNIST where the numerosity of objects placed in an image must be determined. The experiments called SMNIST for Humans are intended to measure the capacity of the Object File System in humans. In this type of experiment the measurement result is in well agreement with the value known from the cognitive psychology literature. The experiments called SMNIST for Machines serve similar purposes but they investigate existing, well known (but originally developed for other purpose) and under development deep learning computer programs. These measurement results can be interpreted similar to the results from SMNIST for Humans. The main thesis of this paper can be formulated as follows: in machines the image classification artificial neural networks can learn to distinguish numerosities with better accuracy when these numerosities are smaller than the capacity of OFS in humans. Finally, we outline a conceptual framework to investigate the notion of number in humans and machines.

Policy gradient based reinforcement learning algorithms coupled with neural networks have shown success in learning complex policies in the model free continuous action space control setting. However, explicitly parameterized policies are limited by the scope of the chosen parametric probability distribution. We show that alternatively to the likelihood based policy gradient, a related objective can be optimized through advantage weighted quantile regression. Our approach models the policy implicitly in the network, which gives the agent the freedom to approximate any distribution in each action dimension, not limiting its capabilities to the commonly used unimodal Gaussian parameterization. This broader spectrum of policies makes our algorithm suitable for problems where Gaussian policies cannot fit the optimal policy. Moreover, our results on the MuJoCo physics simulator benchmarks are comparable or superior to state-of-the-art on-policy methods.