Humans excel at continually acquiring and fine-tuning knowledge over sustained time spans. This ability, typically referred to as lifelong learning, is crucial for artificial agents interacting in real-world, dynamic environments where i) the number of tasks to be learned is not pre-defined, ii) training samples become progressively available over time, and iii) annotated samples may be very sparse. In this paper, we propose a dual-memory self-organizing system that learns spatiotemporal representations from videos. The architecture draws inspiration from the interplay of the hippocampal and neocortical systems in the mammalian brain argued to mediate the complementary tasks of quickly integrating specific experiences, i.e., episodic memory (EM), and slowly learning generalities from episodic events, i.e., semantic memory (SM). The complementary memories are modeled as recurrent self-organizing neural networks: The EM quickly adapts to incoming novel sensory observations via competitive Hebbian Learning, whereas the SM progressively learns compact representations by using task-relevant signals to regulate intrinsic levels of neurogenesis and neuroplasticity. For the consolidation of knowledge, trajectories of neural reactivations are periodically replayed to both networks. We analyze and evaluate the performance of our approach with the CORe50 benchmark dataset for continuous object recognition from videos. We show that the proposed approach significantly outperforms current (supervised) methods of lifelong learning in three different incremental learning scenarios, and that due to the unsupervised nature of neural network self-organization, our approach can be used in scenarios where sample annotations are sparse.

A central area of research in nonlinear science is the study of instabilities that drive the emergence of extreme events. Unfortunately, experimental techniques for measuring such phenomena often provide only partial characterization. For example, real-time studies of instabilities in nonlinear fibre optics frequently use only spectral data, precluding detailed predictions about the associated temporal properties. Here, we show how Machine Learning can overcome this limitation by predicting statistics for the maximum intensity of temporal peaks in modulation instability based only on spectral measurements. Specifically, we train a neural network based Machine Learning model to correlate spectral and temporal properties of optical fibre modulation instability using data from numerical simulations, and we then use this model to predict the temporal probability distribution based on high-dynamic range spectral data from experiments. These results open novel perspectives in all systems exhibiting chaos and instability where direct time-domain observations are difficult.

Semi-implicit variational inference (SIVI) is introduced to expand the commonly used analytic variational distribution family, by mixing the variational parameter with a flexible distribution. This mixing distribution can assume any density function, explicit or not, as long as independent random samples can be generated via reparameterization. Not only does SIVI expand the variational family to incorporate highly flexible variational distributions, including implicit ones that have no analytic density functions, but also sandwiches the evidence lower bound (ELBO) between a lower bound and an upper bound, and further derives an asymptotically exact surrogate ELBO that is amenable to optimization via stochastic gradient ascent. With a substantially expanded variational family and a novel optimization algorithm, SIVI is shown to closely match the accuracy of MCMC in inferring the posterior in a variety of Bayesian inference tasks.

In this paper, we introduce an unsupervised learning approach to automatically discover, summarize, and manipulate artistic styles from large collections of paintings. Our method is based on archetypal analysis, which is an unsupervised learning technique akin to sparse coding with a geometric interpretation. When applied to deep image representations from a collection of artworks, it learns a dictionary of archetypal styles, which can be easily visualized. After training the model, the style of a new image, which is characterized by local statistics of deep visual features, is approximated by a sparse convex combination of archetypes. This enables us to interpret which archetypal styles are present in the input image, and in which proportion. Finally, our approach allows us to manipulate the coefficients of the latent archetypal decomposition, and achieve various special effects such as style enhancement, transfer, and interpolation between multiple archetypes.

We consider an online influence maximization problem in which a decision maker selects a node among a large number of possibilities and places a piece of information at the node. The node transmits the information to some others that are in the same connected component in a random graph. The goal of the decision maker is to reach as many nodes as possible, with the added complication that feedback is only available about the degree of the selected node. Our main result shows that such local observations can be sufficient for maximizing global influence in two broadly studied families of random graph models: stochastic block models and Chung--Lu models. With this insight, we propose a bandit algorithm that aims at maximizing local (and thus global) influence, and provide its theoretical analysis in both the subcritical and supercritical regimes of both considered models. Notably, our performance guarantees show no explicit dependence on the total number of nodes in the network, making our approach well-suited for large-scale applications.

Deep Neural Networks have been applied very successfully in image recognition and natural language processing. Recently these powerful methods have received attention also in the brain-computer interface (BCI) community. Here, we introduce a convolutional neural network (CNN) architecture optimized for classification of brain states from non-invasive magnetoencephalographic (MEG) measurements. The model structure is motivated by a state-of-the-art generative model of the MEG signal and is thus readily interpretable in neurophysiological terms. We demonstrate that the proposed model is highly accurate in decoding event-related responses as well as modulations of oscillatory brain activity, and is robust with respect to inter-individual differences. Importantly, the model generalizes well across users: when trained on data pooled from previous users, it can successfully perform on new users. Thus, the time-consuming BCI calibration can be omitted. Moreover, the model can be incrementally updated, resulting in +8.9% average accuracy improvement in offline experiments and +17.0% in a real-time BCI. We argue that this model can be used in practical BCIs and basic neuroscience research.

Each training step for a variational autoencoder (VAE) requires us to sample from the approximate posterior, so we usually choose simple (e.g. factorised) approximate posteriors in which sampling is an efficient computation that fully exploits GPU parallelism. However, such simple approximate posteriors are often insufficient, as they eliminate statistical dependencies in the posterior. While it is possible to use normalizing flow approximate posteriors for continuous latents, some problems have discrete latents and strong statistical dependencies. The most natural approach to model these dependencies is an autoregressive distribution, but sampling from such distributions is inherently sequential and thus slow. We develop a fast, parallel sampling procedure for autoregressive distributions based on fixed-point iterations which enables efficient and accurate variational inference in discrete state-space latent variable dynamical systems. To optimize the variational bound, we considered two ways to evaluate probabilities: inserting the relaxed samples directly into the pmf for the discrete distribution, or converting to continuous logistic latent variables and interpreting the K-step fixed-point iterations as a normalizing flow. We found that converting to continuous latent variables gave considerable additional scope for mismatch between the true and approximate posteriors, which resulted in biased inferences, we thus used the former approach. Using our fast sampling procedure, we were able to realize the benefits of correlated posteriors, including accurate uncertainty estimates for one cell, and accurate connectivity estimates for multiple cells, in an order of magnitude less time.

A conventional wisdom in statistical learning is that large models require strong regularization to prevent overfitting. This rule has been recently challenged by deep neural networks: despite being expressive enough to fit any training set perfectly, they still generalize well. Here we show that the same is true for linear regression in the under-determined $n\ll p$ situation, provided that one uses the minimum-norm estimator. The case of linear model with least squares loss allows full and exact mathematical analysis. We prove that augmenting a model with many random covariates with small constant variance and using minimum-norm estimator is asymptotically equivalent to adding the ridge penalty. Using toy example simulations as well as real-life high-dimensional data sets, we demonstrate that explicit ridge penalty often fails to provide any improvement over this implicit ridge regularization. In this regime, minimum-norm estimator achieves zero training error but nevertheless has low expected error.

In this work, we propose an end-to-end block-based auto-encoder system for image compression. We introduce novel contributions to neural-network based image compression, mainly in achieving binarization simulation, variable bit rates with multiple networks, entropy-friendly representations, inference-stage code optimization and performance-improving normalization layers in the auto-encoder. We evaluate and show the incremental performance increase of each of our contributions.

Methods in Economics TU Vienna In this work we introduce a novel paradigm for Bayesian learning based on optimal transport theory. Namely, we propose to use the Wasserstein barycenter of the posterior law on models, as an alternative to the maximum a posteriori estimator (MAP) and Bayes predictive distributions. We exhibit conditions granting the existence and consistency of this estimator, discuss some of its basic and specific properties, and propose a numerical approximation relying on standard posterior sampling in general finite-dimensional parameter spaces. We thus also contribute to the recent blooming of applications of optimal transport theory in machine learning, beyond the discrete and semidiscrete settings so far considered. Advantages of the proposed estimator are discussed and illustrated with numerical simulations.