OATML

Thirteen papers with OATML members accepted to NeurIPS 2021 main conference. More information in our blog post.

OATML graduate students Aidan Gomez, Jannik Kossen, and Neil Band will be presenting their recent paper Self-Attention Between Datapoints: Going Beyond Individual Input-Output Pairs in Deep Learning that introduces Non-Parametric Transformers at the Stanford Lecture Course ‘CS25: Transformers United’ on November 1, 2021. Professor Yarin Gal, and OATML DPhil students Clare Lyle and Lewis Smith are co-authors on the paper.

The talk will be made available online.

OATML students Jannik Kossen and Neil Band will be presenting their recent paper Self-Attention Between Datapoints: Going Beyond Individual Input-Output Pairs in Deep Learning at Google Research on September 14, 2021. Professor Yarin Gal, and OATML DPhil students Aidan Gomez, Clare Lyle and Lewis Smith are co-authors on the paper.

OATML DPhil student Jannik Kossen gives invited talks at AI Campus Berlin on two recent papers: Self-Attention Between Datapoints: Going Beyond Individual Input-Output Pairs in Deep Learning and Active Testing: Sample-Efficient Model Evaluation. Recordings of are available upon request. Announcements are here and here. Professor Yarin Gal, Dr Tom Rainforth, and OATML DPhil students Sebastian Farquhar, Aidan Gomez, Clare Lyle and Lewis Smith are co-authors on the papers.

Seven papers with OATML members accepted to ICML 2021, together with 14 workshop papers. More information in our blog post.

OATML students Jannik Kossen and Neil Band present their recent paper Self-Attention Between Datapoints: Going Beyond Individual Input-Output Pairs in Deep Learning at Cohere on July 9, 2021. Professor Yarin Gal, and OATML DPhil students Aidan Gomez, Clare Lyle and Lewis Smith are also co-authors on the paper.

OATML graduate student Clare Lyle will be giving a talk about her work with Professor Yarin Gal on causal inference and generalization in deep reinforcement learning at the Simons Institute Workshop on Deep Reinforcement Learning on Thursday October 1.

OATML graduate student Clare Lyle has been selected for the 2020 Open Philanthropy AI Fellowship. The fellowship will support her research for five years.

We challenge a common assumption underlying most supervised deep learning: that a model makes a prediction depending only on its parameters and the features of a single input. To this end, we introduce a general-purpose deep learning architecture that takes as input the entire dataset instead of processing one datapoint at a time. Our approach uses self-attention to reason about relationships between datapoints explicitly, which can be seen as realizing non-parametric models using parametric attention mechanisms. However, unlike conventional non-parametric models, we let the model learn end-to-end from the data how to make use of other datapoints for prediction. Empirically, our models solve cross-datapoint lookup and complex reasoning tasks unsolvable by traditional deep learning models. We show highly competitive results on tabular data, early results on CIFAR-10, and give insight into how the model makes use of the interactions between points.

Robustness of decision rules to shifts in the data-generating process is crucial to the successful deployment of decision-making systems. Such shifts can be viewed as interventions on a causal graph, which capture (possibly hypothetical) changes in the data-generating process, whether due to natural reasons or by the action of an adversary. We consider causal Bayesian networks and formally define the interventional robustness problem, a novel model-based notion of robustness for decision functions that measures worst-case performance with respect to a set of interventions that denote changes to parameters and/or causal influences. By relying on a tractable representation of Bayesian networks as arithmetic circuits, we provide efficient algorithms for computing guaranteed upper and lower bounds on the interventional robustness probabilities. Experimental results demonstrate that the methods yield useful and interpretable bounds for a range of practical networks, paving the way towar... [full abstract]

We study reinforcement learning (RL) with no-reward demonstrations, a setting in which an RL agent has access to additional data from the interaction of other agents with the same environment. However, it has no access to the rewards or goals of these agents, and their objectives and levels of expertise may vary widely. These assumptions are common in multi-agent settings, such as autonomous driving. To effectively use this data, we turn to the framework of successor features. This allows us to disentangle shared features and dynamics of the environment from agent-specific rewards and policies. We propose a multi-task inverse reinforcement learning (IRL) algorithm, called _inverse temporal difference learning_ (ITD), that learns shared state features, alongside per-agent successor features and preference vectors, purely from demonstrations without reward labels. We further show how to seamlessly integrate ITD with learning from online environment interactions, arriving at a novel a... [full abstract]

We take a Bayesian perspective to illustrate a connection between training speed and the marginal likelihood in linear models. This provides two major insights: first, that a measure of a model's training speed can be used to estimate its marginal likelihood. Second, that this measure, under certain conditions, predicts the relative weighting of models in linear model combinations trained to minimize a regression loss. We verify our results in model selection tasks for linear models and for the infinite-width limit of deep neural networks. We further provide encouraging empirical evidence that the intuition developed in these settings also holds for deep neural networks trained with stochastic gradient descent. Our results suggest a promising new direction towards explaining why neural networks trained with stochastic gradient descent are biased towards functions that generalize well.

The Information Bottleneck principle offers both a mechanism to explain how deep neural networks train and generalize, as well as a regularized objective with which to train models. However, multiple competing objectives are proposed in the literature, and the information-theoretic quantities used in these objectives are difficult to compute for large deep neural networks, which in turn limits their use as a training objective. In this work, we review these quantities and compare and unify previously proposed objectives, which allows us to develop surrogate objectives more friendly to optimization without relying on cumbersome tools such as density estimation. We find that these surrogate objectives allow us to apply the information bottleneck to modern neural network architectures. We demonstrate our insights on MNIST, CIFAR-10 and Imagenette with modern DNN architectures (ResNets).

The Information Bottleneck (IB) principle characterizes learning and generalization in deep neural networks in terms of the change in two information theoretic quantities and leads to a regularized objective function for training neural networks. These quantities are difficult to compute directly for deep neural networks. We show that it is possible to backpropagate through a simple entropy estimator to obtain an IB training method that works for modern neural network architectures. We evaluate our approach empirically on the CIFAR-10 dataset, showing that IB objectives can yield competitive performance on this dataset with a conceptually simple approach while also performing well against adversarial attacks out-of-the-box.

Generalization across environments is critical to the successful application of reinforcement learning algorithms to real-world challenges. In this paper, we consider the problem of learning abstractions that generalize in block MDPs, families of environments with a shared latent state space and dynamics structure over that latent space, but varying observations. We leverage tools from causal inference to propose a method of invariant prediction to learn model-irrelevance state abstractions (MISA) that generalize to novel observations in the multi-environment setting. We prove that for certain classes of environments, this approach outputs with high probability a state abstraction corresponding to the causal feature set with respect to the return. We further provide more general bounds on model error and generalization error in the multi-environment setting, in the process showing a connection between causal variable selection and the state abstraction framework for MDPs. We give e... [full abstract]

Invariance is widely described as a desirable property of neural networks, but the mechanisms by which it benefits deep learning remain shrouded in mystery. We show that building invariance into model architecture via feature averaging provably tightens PAC-Bayes generalization bounds, as compared to data augmentation. Furthermore, through a link to the marginal likelihood and Bayesian model selection, we provide justification for using the improvement in these bounds for model selection. Our key observation is that invariance doesn't just reduce variance in deep learning: it also changes the parameter-function mapping, and this leads better provable guarantees for the model. We verify our theoretical results empirically on a permutation-invariant dataset.

We propose a new perspective on representation learning in reinforcement learning based on geometric properties of the space of value functions. We leverage this perspective to provide formal evidence regarding the usefulness of value functions as auxiliary tasks. Our formulation considers adapting the representation to minimize the (linear) approximation of the value function of all stationary policies for a given environment. We show that this optimization reduces to making accurate predictions regarding a special class of value functions which we call adversarial value functions (AVFs). We demonstrate that using value functions as auxiliary tasks corresponds to an expected-error relaxation of our formulation, with AVFs a natural candidate, and identify a close relationship with proto-value functions (Mahadevan, 2005). We highlight characteristics of AVFs and their usefulness as auxiliary tasks in a series of experiments on the four-room domain.

Invariance is often cited as a desirable property of machine learning systems, claimed to improve model accuracy and reduce overfitting. Empirically, invariant models often generalize better than their non-invariant counterparts. But is it possible to show that invariant models provably do so? In this paper we explore the effect of invariance on model generalization. We find strong Bayesian and frequentist motivations for enforcing invariance which leverage recent results connecting PAC-Bayes generalization bounds and the marginal likelihood. We make use of these results to perform model selection on neural networks.

Since their introduction a year ago, distributional approaches to reinforcement learning (distributional RL) have produced strong results relative to the standard approach which models expected values (expected RL). However, aside from convergence guarantees, there have been few theoretical results investigating the reasons behind the improvements distributional RL provides. In this paper we begin the investigation into this fundamental question by analyzing the differences in the tabular, linear approximation, and non-linear approximation settings. We prove that in many realizations of the tabular and linear approximation settings, distributional RL behaves exactly the same as expected RL. In cases where the two methods behave differently, distributional RL can in fact hurt performance when it does not induce identical behaviour. We then continue with an empirical analysis comparing distributional and expected RL methods in control settings with non-linear approximators to tease a... [full abstract]

OATML group members and collaborators are proud to present 13 papers at NeurIPS 2021 main conference. …

OATML group members and collaborators are proud to present 21 papers at ICML 2021, including 7 papers at the main conference and 14 papers at various workshops. Group members will also be giving invited talks and participate in panel discussions at the workshops. …

OATML group members and collaborators are proud to be presenting 22 papers at NeurIPS 2020. Group members are also co-organising various events around NeurIPS, including workshops, the NeurIPS Meet-Up on Bayesian Deep Learning and socials. …

We are glad to share the following 13 papers by OATML authors and collaborators to be presented at this ICML conference and workshops …

We are glad to share the following 25 papers by OATML authors and collaborators to be presented at this NeurIPS conference and workshops. …

The applications of probably approximately correct (PAC) learning results to deep networks have historically been about as interesting as they sound. For neural networks of the scale used in practical applications, bounds involving concepts like VC dimension conclude that the algorithm will have no more than a certain error rate on the test set with probability at least zero. Recently, some work by Dziugaite and Roy, along with some folks from Columbia has managed to obtain non-vacuous generalization bounds for more realistic problems using a concept introduced by McAllester (1999) called PAC Bayes bounds. …