Training generative adversarial networks requires balancing of delicate adversarial dynamics. Even with careful tuning, training may diverge or end up in a bad equilibrium with dropped modes. In this work, we introduce a new form of latent optimisation inspired by the CS-GAN and show that it improves adversarial dynamics by enhancing interactions between the discriminator and the generator. We develop supporting theoretical analysis from the perspectives of differentiable games and stochastic approximation. Our experiments demonstrate that latent optimisation can significantly improve GAN training, obtaining state-of-the-art performance for the ImageNet ( 128 128) dataset. Our model achieves an Inception Score (IS) of 148 and an Fr echet Inception Distance (FID) of 3.4, an improvement of 17% and 32% in IS and FID respectively, compared with the baseline BigGAN-deep model with the same architecture and number of parameters. Generative Adversarial Nets (GANs) are implicit generative models that can be trained to match a given data distribution. GANs were originally developed by Goodfellow et al. (2014) for image data. As the field of generative modelling has advanced, GANs remain at the frontier, generating high-fidelity images at large scale (Brock et al., 2018).

Large-scale non-convex sparsity-constrained problems have recently gained extensive attention. Most existing deterministic optimization methods (e.g., GraSP) are not suitable for large-scale and high-dimensional problems, and thus stochastic optimization methods with hard thresholding (e.g., SVRGHT) become more attractive. Inspired by GraSP, this paper proposes a new general relaxed gradient support pursuit (RGraSP) framework, in which the sub-algorithm only requires to satisfy a slack descent condition. We also design two specific semi-stochastic gradient hard thresholding algorithms. In particular, our algorithms have much less hard thresholding operations than SVRGHT, and their average per-iteration cost is much lower (i.e., O(d) vs. O(d log(d)) for SVRGHT), which leads to faster convergence. Our experimental results on both synthetic and real-world datasets show that our algorithms are superior to the state-of-the-art gradient hard thresholding methods.

Bipartite ranking is an important supervised learning problem; however, unlike regression or classification, it has a quadratic dependence on the number of samples. To circumvent the prohibitive sample cost, many recent work focus on stochastic gradient-based methods. In this paper we consider an alternative approach, which leverages the structure of the widely-adopted pairwise squared loss, to obtain a stochastic and low cost algorithm that does not require stochastic gradients or learning rates. Using a novel uniform risk bound---based on matrix and vector concentration inequalities---we show that the sample size required for competitive performance against the all-pairs batch algorithm does not have a quadratic dependence. Generalization bounds for both the batch and low cost stochastic algorithms are presented. Experimental results show significant speed gain against the batch algorithm, as well as competitive performance against state-of-the-art bipartite ranking algorithms on real datasets.

Quantum computing has received significant attention as a next-generation computing technology due to its potential speed and ability to solve problems considered too difficult for classical computers, as reflected in the recent discussion on Quantum Supremacy. Grid sees quantum computing not only as a tool for solving optimization and quantum chemical computation problems, but also as a tool for AI (Machine Learning, Deep Learning, etc.) calculations, such as feature extraction. Previous works have announced the successful implementation of machine learning-related algorithms, such as principal component analysis and auto-encoders, on quantum computers. This work announces the development of a gradient descent (backpropagation) algorithm, a method commonly used in machine learning for neural network parameter optimization, for use on NISQ quantum computers. Due to the non-linear nature of quantum bits (qubits), Grid proposes that this algorithm can be used to perform the feature extraction and representation calculations that deep learning methods employ.

Stochastic learning control of inhomogeneous quantum ensembles Gabriel Turinici IUF - Institut Universitaire de France CEREMADE, Universit e Paris Dauphine - PSL Research University Oct 2019 Abstract In quantum control, the robustness with respect to uncertainties in the system's parameters or driving field characteristics is of paramount importance and has been studied theoretically, numerically and experimentally. We test in this paper stochastic search procedures (Stochastic gradient descent and the Adam algorithm) that sample, at each iteration, from the distribution of the parameter uncertainty, as opposed to previous approaches that use a fixed grid. We show that both algorithms behave well with respect to benchmarks and discuss their relative merits. In addition the methodology allows to address high dimensional parameter uncertainty; we implement numerically, with good results, a 3D and a 6D case. 1 Introduction Quantum control is a promising technology with many applications ranging from NMR [12] to quantum computing [15] and laser control of quantum dynamics [7]. The controlling field encounters many molecules which although identical in nature may interact differently with the incoming field because of e.g., different Larmor frequencies or rf attenuation factors (in NMR spin control or quantum computing, see [19, 29, 35, 22, 13, 17]), different spatial profile (see [24]) or other parameters (see [36, 8, 10]). For obvious practical reasons, it is of paramount importance to ensure that the control quality is 1 arXiv:1906.02991v3

The gradient noise (GN) in the stochastic gradient descent (SGD) algorithm is often considered to be Gaussian in the large data regime by assuming that the \emph{classical} central limit theorem (CLT) kicks in. This assumption is often made for mathematical convenience, since it enables SGD to be analyzed as a stochastic differential equation (SDE) driven by a Brownian motion. We argue that the Gaussianity assumption might fail to hold in deep learning settings and hence render the Brownian motion-based analyses inappropriate. Inspired by non-Gaussian natural phenomena, we consider the GN in a more general context and invoke the \emph{generalized} CLT, which suggests that the GN converges to a \emph{heavy-tailed} $\alpha$-stable random vector, where \emph{tail-index} $\alpha$ determines the heavy-tailedness of the distribution. Accordingly, we propose to analyze SGD as a discretization of an SDE driven by a L\'{e}vy motion. Such SDEs can incur `jumps', which force the SDE and its discretization \emph{transition} from narrow minima to wider minima, as proven by existing metastability theory and the extensions that we proved recently. In this study, under the $\alpha$-stable GN assumption, we further establish an explicit connection between the convergence rate of SGD to a local minimum and the tail-index $\alpha$. To validate the $\alpha$-stable assumption, we conduct experiments on common deep learning scenarios and show that in all settings, the GN is highly non-Gaussian and admits heavy-tails. We investigate the tail behavior in varying network architectures and sizes, loss functions, and datasets. Our results open up a different perspective and shed more light on the belief that SGD prefers wide minima.

In few-shot learning, typically, the loss function which is applied at test time is the one we are ultimately interested in minimising, such as the mean-squared-error loss for a regression problem. However, given that we have few samples at test time, we argue that the loss function that we are interested in minimising is not necessarily the loss function most suitable for computing gradients in a few-shot setting. We propose VIABLE, a generic meta-learning extension that builds on existing meta-gradient-based methods by learning a differentiable loss function, replacing the pre-defined inner-loop loss function in performing task-specific updates. We show that learning a loss function capable of leveraging relational information between samples reduces underfitting, and significantly improves performance and sample efficiency on a simple regression task. Furthermore, we show VIABLE is scalable by evaluating on the Mini-Imagenet dataset.

A remarkable recent discovery in machine learning has been that deep neural networks can achieve impressive performance (in terms of both lower training error and higher generalization capacity) in the regime where they are massively over-parameterized. Consequently, over the last several months, the community has devoted growing interest in analyzing optimization and generalization properties of over-parameterized networks, and several breakthrough works have led to important theoretical progress. However, the majority of existing work only applies to supervised learning scenarios and hence are limited to settings such as classification and regression. In contrast, the role of over-parameterization in the unsupervised setting has gained far less attention. In this paper, we study the gradient dynamics of two-layer over-parameterized autoencoders with ReLU activation. We make very few assumptions about the given training dataset (other than mild non-degeneracy conditions). Starting from a randomly initialized autoencoder network, we rigorously prove the linear convergence of gradient descent in two learning regimes, namely: (i) the weakly-trained regime where only the encoder is trained, and (ii) the jointly-trained regime where both the encoder and the decoder are trained. Our results indicate the considerable benefits of joint training over weak training for finding global optima, achieving a dramatic decrease in the required level of over-parameterization. We also analyze the case of weight-tied autoencoders (which is a commonly used architectural choice in practical settings) and prove that in the over-parameterized setting, training such networks from randomly initialized points leads to certain unexpected degeneracies.

Motivated by the flexibility of biological neural networks whose connectivity structure changes significantly during their lifetime, we introduce the Unstructured Recursive Network (URN) and demonstrate that it can exhibit similar flexibility during training via gradient descent. We show empirically that many of the different neural network structures commonly used in practice today (including fully connected, locally connected and residual networks of different depths and widths) can emerge dynamically from the same URN. These different structures can be derived using gradient descent on a single general loss function where the structure of the data and the relative strengths of various regulator terms determine the structure of the emergent network. We show that this loss function and the regulators arise naturally when considering the symmetries of the network as well as the geometric properties of the input data.