Sunil Gupta

Thanh Nguyen-Tang, Ming Yin, Sunil Gupta et al. · princeton

Sample-efficient offline reinforcement learning (RL) with linear function approximation has recently been studied extensively. Much of prior work has yielded the minimax-optimal bound of $\tilde{\mathcal{O}}(\frac{1}{\sqrt{K}})$, with $K$ being the number of episodes in the offline data. In this work, we seek to understand instance-dependent bounds for offline RL with function approximation. We present an algorithm called Bootstrapped and Constrained Pessimistic Value Iteration (BCP-VI), which leverages data bootstrapping and constrained optimization on top of pessimism. We show that under a partial data coverage assumption, that of \emph{concentrability} with respect to an optimal policy, the proposed algorithm yields a fast rate of $\tilde{\mathcal{O}}(\frac{1}{K})$ for offline RL when there is a positive gap in the optimal Q-value functions, even when the offline data were adaptively collected. Moreover, when the linear features of the optimal actions in the states reachable by an optimal policy span those reachable by the behavior policy and the optimal actions are unique, offline RL achieves absolute zero sub-optimality error when $K$ exceeds a (finite) instance-dependent threshold. To the best of our knowledge, these are the first $\tilde{\mathcal{O}}(\frac{1}{K})$ bound and absolute zero sub-optimality bound respectively for offline RL with linear function approximation from adaptive data with partial coverage. We also provide instance-agnostic and instance-dependent information-theoretical lower bounds to complement our upper bounds.

Hung Tran-The, Sunil Gupta, Santu Rana et al.

The expected improvement (EI) algorithm is one of the most popular strategies for optimization under uncertainty due to its simplicity and efficiency. Despite its popularity, the theoretical aspects of this algorithm have not been properly analyzed. In particular, whether in the noisy setting, the EI strategy with a standard incumbent converges is still an open question of the Gaussian process bandit optimization problem. We aim to answer this question by proposing a variant of EI with a standard incumbent defined via the GP predictive mean. We prove that our algorithm converges, and achieves a cumulative regret bound of $\mathcal O(γ_T\sqrt{T})$, where $γ_T$ is the maximum information gain between $T$ observations and the Gaussian process model. Based on this variant of EI, we further propose an algorithm called Improved GP-EI that converges faster than previous counterparts. In particular, our proposed variants of EI do not require the knowledge of the RKHS norm and the noise's sub-Gaussianity parameter as in previous works. Empirical validation in our paper demonstrates the effectiveness of our algorithms compared to several baselines.

Dupati Srikar Chandra, Sakshi Varshney, P. K. Srijith et al.

Humans learn continually throughout their lifespan by accumulating diverse knowledge and fine-tuning it for future tasks. When presented with a similar goal, neural networks suffer from catastrophic forgetting if data distributions across sequential tasks are not stationary over the course of learning. An effective approach to address such continual learning (CL) problems is to use hypernetworks which generate task dependent weights for a target network. However, the continual learning performance of existing hypernetwork based approaches are affected by the assumption of independence of the weights across the layers in order to maintain parameter efficiency. To address this limitation, we propose a novel approach that uses a dependency preserving hypernetwork to generate weights for the target network while also maintaining the parameter efficiency. We propose to use recurrent neural network (RNN) based hypernetwork that can generate layer weights efficiently while allowing for dependencies across them. In addition, we propose novel regularisation and network growth techniques for the RNN based hypernetwork to further improve the continual learning performance. To demonstrate the effectiveness of the proposed methods, we conducted experiments on several image classification continual learning tasks and settings. We found that the proposed methods based on the RNN hypernetworks outperformed the baselines in all these CL settings and tasks.

Thao Minh Le, Vuong Le, Sunil Gupta et al.

The current success of modern visual reasoning systems is arguably attributed to cross-modality attention mechanisms. However, in deliberative reasoning such as in VQA, attention is unconstrained at each step, and thus may serve as a statistical pooling mechanism rather than a semantic operation intended to select information relevant to inference. This is because at training time, attention is only guided by a very sparse signal (i.e. the answer label) at the end of the inference chain. This causes the cross-modality attention weights to deviate from the desired visual-language bindings. To rectify this deviation, we propose to guide the attention mechanism using explicit linguistic-visual grounding. This grounding is derived by connecting structured linguistic concepts in the query to their referents among the visual objects. Here we learn the grounding from the pairing of questions and images alone, without the need for answer annotation or external grounding supervision. This grounding guides the attention mechanism inside VQA models through a duality of mechanisms: pre-training attention weight calculation and directly guiding the weights at inference time on a case-by-case basis. The resultant algorithm is capable of probing attention-based reasoning models, injecting relevant associative knowledge, and regulating the core reasoning process. This scalable enhancement improves the performance of VQA models, fortifies their robustness to limited access to supervised data, and increases interpretability.

Sunil Gupta, Alistair Shilton, Arun Kumar A et al.

In this paper we introduce BO-Muse, a new approach to human-AI teaming for the optimization of expensive black-box functions. Inspired by the intrinsic difficulty of extracting expert knowledge and distilling it back into AI models and by observations of human behavior in real-world experimental design, our algorithm lets the human expert take the lead in the experimental process. The human expert can use their domain expertise to its full potential, while the AI plays the role of a muse, injecting novelty and searching for areas of weakness to break the human out of over-exploitation induced by cognitive entrenchment. With mild assumptions, we show that our algorithm converges sub-linearly, at a rate faster than the AI or human alone. We validate our algorithm using synthetic data and with human experts performing real-world experiments.

Hung Le, Thommen Karimpanal George, Majid Abdolshah et al.

We introduce a constrained optimization method for policy gradient reinforcement learning, which uses a virtual trust region to regulate each policy update. In addition to using the proximity of one single old policy as the normal trust region, we propose forming a second trust region through another virtual policy representing a wide range of past policies. We then enforce the new policy to stay closer to the virtual policy, which is beneficial if the old policy performs poorly. More importantly, we propose a mechanism to automatically build the virtual policy from a memory of past policies, providing a new capability for dynamically learning appropriate virtual trust regions during the optimization process. Our proposed method, dubbed Memory-Constrained Policy Optimization (MCPO), is examined in diverse environments, including robotic locomotion control, navigation with sparse rewards and Atari games, consistently demonstrating competitive performance against recent on-policy constrained policy gradient methods.

Phuoc Nguyen, Truyen Tran, Ky Le et al.

We introduce a conditional compression problem and propose a fast framework for tackling it. The problem is how to quickly compress a pretrained large neural network into optimal smaller networks given target contexts, e.g. a context involving only a subset of classes or a context where only limited compute resource is available. To solve this, we propose an efficient Bayesian framework to compress a given large network into much smaller size tailored to meet each contextual requirement. We employ a hypernetwork to parameterize the posterior distribution of weights given conditional inputs and minimize a variational objective of this Bayesian neural network. To further reduce the network sizes, we propose a new input-output group sparsity factorization of weights to encourage more sparseness in the generated weights. Our methods can quickly generate compressed networks with significantly smaller sizes than baseline methods.

Alistair Shilton, Sunil Gupta, Santu Rana et al.

The study of Neural Tangent Kernels (NTKs) has provided much needed insight into convergence and generalization properties of neural networks in the over-parametrized (wide) limit by approximating the network using a first-order Taylor expansion with respect to its weights in the neighborhood of their initialization values. This allows neural network training to be analyzed from the perspective of reproducing kernel Hilbert spaces (RKHS), which is informative in the over-parametrized regime, but a poor approximation for narrower networks as the weights change more during training. Our goal is to extend beyond the limits of NTK toward a more general theory. We construct an exact power-series representation of the neural network in a finite neighborhood of the initial weights as an inner product of two feature maps, respectively from data and weight-step space, to feature space, allowing neural network training to be analyzed from the perspective of reproducing kernel {\em Banach} space (RKBS). We prove that, regardless of width, the training sequence produced by gradient descent can be exactly replicated by regularized sequential learning in RKBS. Using this, we present novel bound on uniform convergence where the iterations count and learning rate play a central role, giving new theoretical insight into neural network training.

Dat Phan-Trong, Hung Tran-The, Sunil Gupta

Bayesian Optimization (BO) is an effective approach for global optimization of black-box functions when function evaluations are expensive. Most prior works use Gaussian processes to model the black-box function, however, the use of kernels in Gaussian processes leads to two problems: first, the kernel-based methods scale poorly with the number of data points and second, kernel methods are usually not effective on complex structured high dimensional data due to curse of dimensionality. Therefore, we propose a novel black-box optimization algorithm where the black-box function is modeled using a neural network. Our algorithm does not need a Bayesian neural network to estimate predictive uncertainty and is therefore computationally favorable. We analyze the theoretical behavior of our algorithm in terms of regret bound using advances in NTK theory showing its efficient convergence. We perform experiments with both synthetic and real-world optimization tasks and show that our algorithm is more sample efficient compared to existing methods.

Haripriya Harikumar, Santu Rana, Kien Do et al.

Adversarial attacks on deep learning-based models pose a significant threat to the current AI infrastructure. Among them, Trojan attacks are the hardest to defend against. In this paper, we first introduce a variation of the Badnet kind of attacks that introduces Trojan backdoors to multiple target classes and allows triggers to be placed anywhere in the image. The former makes it more potent and the latter makes it extremely easy to carry out the attack in the physical space. The state-of-the-art Trojan detection methods fail with this threat model. To defend against this attack, we first introduce a trigger reverse-engineering mechanism that uses multiple images to recover a variety of potential triggers. We then propose a detection mechanism by measuring the transferability of such recovered triggers. A Trojan trigger will have very high transferability i.e. they make other images also go to the same class. We study many practical advantages of our attack method and then demonstrate the detection performance using a variety of image datasets. The experimental results show the superior detection performance of our method over the state-of-the-arts.

Manisha Senadeera, Thommen Karimpanal George, Sunil Gupta et al.

We can usually assume others have goals analogous to our own. This assumption can also, at times, be applied to multi-agent games - e.g. Agent 1's attraction to green pellets is analogous to Agent 2's attraction to red pellets. This "analogy" assumption is tied closely to the cognitive process known as empathy. Inspired by empathy, we design a simple and explainable architecture to model another agent's action-value function. This involves learning an "Imagination Network" to transform the other agent's observed state in order to produce a human-interpretable "empathetic state" which, when presented to the learning agent, produces behaviours that mimic the other agent. Our approach is applicable to multi-agent scenarios consisting of a single learning agent and other (independent) agents acting according to fixed policies. This architecture is particularly beneficial for (but not limited to) algorithms using a composite value or reward function. We show our method produces better performance in multi-agent games, where it robustly estimates the other's model in different environment configurations. Additionally, we show that the empathetic states are human interpretable, and thus verifiable.

Bao Duong, Sunil Gupta, Thin Nguyen

Existing score-based methods for directed acyclic graph (DAG) learning from observational data struggle to recover the causal graph accurately and sample-efficiently. To overcome this, in this study, we propose DrBO (DAG recovery via Bayesian Optimization)-a novel DAG learning framework leveraging Bayesian optimization (BO) to find high-scoring DAGs. We show that, by sophisticatedly choosing the promising DAGs to explore, we can find higher-scoring ones much more efficiently. To address the scalability issues of conventional BO in DAG learning, we replace Gaussian Processes commonly employed in BO with dropout neural networks, trained in a continual manner, which allows for (i) flexibly modeling the DAG scores without overfitting, (ii) incorporation of uncertainty into the estimated scores, and (iii) scaling with the number of evaluations. As a result, DrBO is computationally efficient and can find the accurate DAG in fewer trials and less time than existing state-of-the-art methods. This is demonstrated through an extensive set of empirical evaluations on many challenging settings with both synthetic and real data. Our implementation is available at https://github.com/baosws/DrBO.

Thao Minh Le, Vuong Le, Kien Do et al.

This paper introduces a new problem, Causal Abductive Reasoning on Video Events (CARVE), which involves identifying causal relationships between events in a video and generating hypotheses about causal chains that account for the occurrence of a target event. To facilitate research in this direction, we create two new benchmark datasets with both synthetic and realistic videos, accompanied by trigger-target labels generated through a novel counterfactual synthesis approach. To explore the challenge of solving CARVE, we present a Causal Event Relation Network (CERN) that examines the relationships between video events in temporal and semantic spaces to efficiently determine the root-cause trigger events. Through extensive experiments, we demonstrate the critical roles of event relational representation learning and interaction modeling in solving video causal reasoning challenges. The introduction of the CARVE task, along with the accompanying datasets and the CERN framework, will advance future research on video causal reasoning and significantly facilitate various applications, including video surveillance, root-cause analysis and movie content management.

Dat Phan-Trong, Hung The Tran, Alistair Shilton et al.

Black-box optimization is a powerful approach for discovering global optima in noisy and expensive black-box functions, a problem widely encountered in real-world scenarios. Recently, there has been a growing interest in leveraging domain knowledge to enhance the efficacy of machine learning methods. Partial Differential Equations (PDEs) often provide an effective means for elucidating the fundamental principles governing the black-box functions. In this paper, we propose PINN-BO, a black-box optimization algorithm employing Physics-Informed Neural Networks that integrates the knowledge from Partial Differential Equations (PDEs) to improve the sample efficiency of the optimization. We analyze the theoretical behavior of our algorithm in terms of regret bound using advances in NTK theory and prove that the use of the PDE alongside the black-box function evaluations, PINN-BO leads to a tighter regret bound. We perform several experiments on a variety of optimization tasks and show that our algorithm is more sample-efficient compared to existing methods.

Thommen George Karimpanal, Hung Le, Majid Abdolshah et al.

The optimistic nature of the Q-learning target leads to an overestimation bias, which is an inherent problem associated with standard $Q-$learning. Such a bias fails to account for the possibility of low returns, particularly in risky scenarios. However, the existence of biases, whether overestimation or underestimation, need not necessarily be undesirable. In this paper, we analytically examine the utility of biased learning, and show that specific types of biases may be preferable, depending on the scenario. Based on this finding, we design a novel reinforcement learning algorithm, Balanced Q-learning, in which the target is modified to be a convex combination of a pessimistic and an optimistic term, whose associated weights are determined online, analytically. We prove the convergence of this algorithm in a tabular setting, and empirically demonstrate its superior learning performance in various environments.

Majid Abdolshah, Hung Le, Thommen Karimpanal George et al.

Transfer in reinforcement learning is usually achieved through generalisation across tasks. Whilst many studies have investigated transferring knowledge when the reward function changes, they have assumed that the dynamics of the environments remain consistent. Many real-world RL problems require transfer among environments with different dynamics. To address this problem, we propose an approach based on successor features in which we model successor feature functions with Gaussian Processes permitting the source successor features to be treated as noisy measurements of the target successor feature function. Our theoretical analysis proves the convergence of this approach as well as the bounded error on modelling successor feature functions with Gaussian Processes in environments with both different dynamics and rewards. We demonstrate our method on benchmark datasets and show that it outperforms current baselines.

Thanh Tang Nguyen, Sunil Gupta, Svetha Venkatesh

We consider the problem of learning a set of probability distributions from the empirical Bellman dynamics in distributional reinforcement learning (RL), a class of state-of-the-art methods that estimate the distribution, as opposed to only the expectation, of the total return. We formulate a method that learns a finite set of statistics from each return distribution via neural networks, as in (Bellemare, Dabney, and Munos 2017; Dabney et al. 2018b). Existing distributional RL methods however constrain the learned statistics to \emph{predefined} functional forms of the return distribution which is both restrictive in representation and difficult in maintaining the predefined statistics. Instead, we learn \emph{unrestricted} statistics, i.e., deterministic (pseudo-)samples, of the return distribution by leveraging a technique from hypothesis testing known as maximum mean discrepancy (MMD), which leads to a simpler objective amenable to backpropagation. Our method can be interpreted as implicitly matching all orders of moments between a return distribution and its Bellman target. We establish sufficient conditions for the contraction of the distributional Bellman operator and provide finite-sample analysis for the deterministic samples in distribution approximation. Experiments on the suite of Atari games show that our method outperforms the standard distributional RL baselines and sets a new record in the Atari games for non-distributed agents.

Julian Berk, Sunil Gupta, Santu Rana et al.

In order to improve the performance of Bayesian optimisation, we develop a modified Gaussian process upper confidence bound (GP-UCB) acquisition function. This is done by sampling the exploration-exploitation trade-off parameter from a distribution. We prove that this allows the expected trade-off parameter to be altered to better suit the problem without compromising a bound on the function's Bayesian regret. We also provide results showing that our method achieves better performance than GP-UCB in a range of real-world and synthetic problems.

Phuoc Nguyen, Truyen Tran, Sunil Gupta et al.

We propose a framework called HyperVAE for encoding distributions of distributions. When a target distribution is modeled by a VAE, its neural network parameters θis drawn from a distribution p(θ) which is modeled by a hyper-level VAE. We propose a variational inference using Gaussian mixture models to implicitly encode the parameters θinto a low dimensional Gaussian distribution. Given a target distribution, we predict the posterior distribution of the latent code, then use a matrix-network decoder to generate a posterior distribution q(θ). HyperVAE can encode the parameters θin full in contrast to common hyper-networks practices, which generate only the scale and bias vectors as target-network parameters. Thus HyperVAE preserves much more information about the model for each task in the latent space. We discuss HyperVAE using the minimum description length (MDL) principle and show that it helps HyperVAE to generalize. We evaluate HyperVAE in density estimation tasks, outlier detection and discovery of novel design classes, demonstrating its efficacy.

Thommen George Karimpanal, Santu Rana, Sunil Gupta et al.

Prior access to domain knowledge could significantly improve the performance of a reinforcement learning agent. In particular, it could help agents avoid potentially catastrophic exploratory actions, which would otherwise have to be experienced during learning. In this work, we identify consistently undesirable actions in a set of previously learned tasks, and use pseudo-rewards associated with them to learn a prior policy. In addition to enabling safer exploratory behaviors in subsequent tasks in the domain, we show that these priors are transferable to similar environments, and can be learned off-policy and in parallel with the learning of other tasks in the domain. We compare our approach to established, state-of-the-art algorithms in both discrete as well as continuous environments, and demonstrate that it exhibits a safer exploratory behavior while learning to perform arbitrary tasks in the domain. We also present a theoretical analysis to support these results, and briefly discuss the implications and some alternative formulations of this approach, which could also be useful in certain scenarios.

Tinu Theckel Joy, Santu Rana, Sunil Gupta et al.

In this paper we develop a Bayesian optimization based hyperparameter tuning framework inspired by statistical learning theory for classifiers. We utilize two key facts from PAC learning theory; the generalization bound will be higher for a small subset of data compared to the whole, and the highest accuracy for a small subset of data can be achieved with a simple model. We initially tune the hyperparameters on a small subset of training data using Bayesian optimization. While tuning the hyperparameters on the whole training data, we leverage the insights from the learning theory to seek more complex models. We realize this by using directional derivative signs strategically placed in the hyperparameter search space to seek a more complex model than the one obtained with small data. We demonstrate the performance of our method on the tasks of tuning the hyperparameters of several machine learning algorithms.