ICML 2023 Papers

In this paper, a unified framework for exploration in reinforcement learning (RL) is proposed based on an option-critic architecture. The proposed framework learns to integrate a set of diverse exploration strategies so that the agent can adaptively select the most effective exploration strategy to realize an effective exploration-exploitation trade-off for each given task. The effectiveness of the proposed exploration framework is demonstrated by various experiments in the MiniGrid and Atari environments.

Dictionary learning, which approximates data samples by a set of shared atoms, is a fundamental task in representation learning. However, dictionary learning over graphs, namely graph dictionary learning (GDL), is much more challenging than vectorial data as graphs lie in disparate metric spaces. The sparse literature on GDL formulates the problem from the reconstructive view and often learns linear graph embeddings with a high computational cost. In this paper, we propose a Fused Gromov-Wasserstein (FGW) Mixture Model named FraMe to address the GDL problem from the generative view. Equipped with the graph generation function based on the radial basis function kernel and FGW distance, FraMe generates nonlinear embedding spaces, which, as we theoretically proved, provide a good approximation of the original graph spaces. A fast solution is further proposed on top of the expectation-maximization algorithm with guaranteed convergence. Extensive experiments demonstrate the effectiveness of the obtained node and graph embeddings, and our algorithm achieves significant improvements over the state-of-the-art methods.

We study distributed contextual linear bandits with stochastic contexts, where $N$ agents/learners act cooperatively to solve a linear bandit-optimization problem with $d$-dimensional features over the course of $T$ rounds. For this problem, we derive the first ever information-theoretic lower bound $\Omega(dN)$ on the communication cost of any algorithm that performs optimally in a regret minimization setup. We then propose a distributed batch elimination version of the LinUCB algorithm, DisBE-LUCB, where the agents share information among each other through a central server. We prove that the communication cost of DisBE-LUCB, matches our lower bound up to logarithmic factors. In particular, for scenarios with known context distribution, the communication cost of DisBE-LUCB is only $\tilde{\mathcal{O}}(dN)$ and its regret is $\tilde{\mathcal{O}}(\sqrt{dNT})$, which is of the same order as that incurred by an optimal single-agent algorithm for $NT$ rounds. We also provide similar bounds for practical settings where the context distribution can only be estimated. Therefore, our proposed algorithm is nearly minimax optimal in terms of *both regret and communication cost*. Finally, we propose DecBE-LUCB, a fully decentralized version of DisBE-LUCB, which operates without a central server, where agents share information with their *immediate neighbors* through a carefully designed consensus procedure.

Temperature scaling is a popular technique for tuning the sharpness of a model distribution. It is used extensively for sampling likely generations and calibrating model uncertainty, and even features as a controllable parameter to many large language models in deployment. However, autoregressive models rely on myopic temperature scaling that greedily optimizes the next token. To address this, we propose *Long Horizon Temperature Scaling* (LHTS), a novel approach for sampling from temperature-scaled *joint* distributions. LHTS is compatible with all likelihood-based models, and optimizes for the long-horizon likelihood of samples. We derive a temperature-dependent LHTS objective, and show that fine-tuning a model on a range of temperatures produces a single model capable of generation with a controllable long-horizon temperature parameter. We experiment with LHTS on image diffusion models and character/language autoregressive models, demonstrating its advantages over myopic temperature scaling in likelihood and sample quality, and showing improvements in accuracy of a multiple choice analogy by $10$%.

Randomized smoothing is a popular method to certify robustness of image classifiers to adversarial input perturbations. It is the only certification technique which scales directly to datasets of higher dimension such as ImageNet. However, current techniques are not able to utilize the fact that any adversarial example has to lie in the image space, that is $[0,1]^d$; otherwise, one can trivially detect it. To address this suboptimality, we derive new certification formulae which lead to significant improvements in the certified $\ell_1$-robustness without the need of adapting the classifiers or change of smoothing distributions. The code is released at https://github.com/vvoracek/L1-smoothing

Homomorphic Encryption (HE) is a promising technology to protect clients' data privacy for Machine Learning as a Service (MLaaS) on public clouds. However, HE operations can be orders of magnitude slower than their counterparts for plaintexts and thus result in prohibitively high inference latency, seriously hindering the practicality of HE. In this paper, we propose a HE-based fast neural network (NN) inference framework--SpENCNN built upon the co-design of HE operation-aware model sparsity and the single-instruction-multiple-data (SIMD)-friendly data packing, to improve NN inference latency. In particular, we first develop an encryption-aware HE-group convolution technique that can partition channels among different groups based on the data size and ciphertext size, and then encode them into the same ciphertext by novel group-interleaved encoding, so as to dramatically reduce the number of bottlenecked operations in HE convolution. We further tailor a HE-friendly sub-block weight pruning to reduce the costly HE-based convolution operation. Our experiments show that SpENCNN can achieve overall speedups of 8.37$\times$, 12.11$\times$, 19.26$\times$, and 1.87$\times$ for LeNet, VGG-5, HEFNet, and ResNet-20 respectively, with negligible accuracy loss. Our code is publicly available at https://github.com/ranran0523/SPECNN.

The goal in episodic memory (EM) is to search a long egocentric video to answer a natural language query (e.g., “where did I leave my purse?”). Existing EM methods exhaustively extract expensive fixed-length clip features to look everywhere in the video for the answer, which is infeasible for long wearable-camera videos that span hours or even days. We propose SpotEM, an approach to achieve efficiency for a given EM method while maintaining good accuracy. SpotEM consists of three key ideas: 1) a novel clip selector that learns to identify promising video regions to search conditioned on the language query; 2) a set of low-cost semantic indexing features that capture the context of rooms, objects, and interactions that suggest where to look; and 3) distillation losses that address the optimization issues arising from end-to-end joint training of the clip selector and EM model. Our experiments on 200+ hours of video from the Ego4D EM Natural Language Queries benchmark and three different EM models demonstrate the effectiveness of our approach: computing only 10% – 25% of the clip features, we preserve 84% – 97% of the original EM model’s accuracy. Project page: https://vision.cs.utexas.edu/projects/spotem

This paper studies the multi-agent performative prediction (Multi-PP) games over multiplex networks. We consider a distributed learning setting where agents partially cooperate on an agent network, while during learning, the data samples drawn depend on the prediction models of the agent itself and neighboring agents on a population network. The dynamics of Multi-PP games is hence affected by the interplay between both networks. This paper concentrates on this Multi-PP game with the following contributions. Firstly, we analyze sufficient conditions for the existence of the performative stable equilibrium (PSE) and Nash equilibrium (NE) of the Multi-PP games. Secondly, we analyze the changes to the equilibrium induced by perturbed data distributions, and derive the closed-form solutions where the network topologies are explicit. Our results connect the existence of PSE/NE with strengths of agents' cooperation, and the changes of equilibrium solutions across agents with their node centrality, etc. Lastly, we show that a stochastic gradient descent (SGD) based distributed learning procedure finds the PSE under the said sufficient condition. Numerical illustrations on the network effects in Multi-PP games corroborate our findings.

Coordinate denoising is a promising 3D molecular pre-training method, which has achieved remarkable performance in various downstream drug discovery tasks. Theoretically, the objective is equivalent to learning the force field, which is revealed helpful for downstream tasks. Nevertheless, there are two challenges for coordinate denoising to learn an effective force field, i.e. low coverage samples and isotropic force field. The underlying reason is that molecular distributions assumed by existing denoising methods fail to capture the anisotropic characteristic of molecules. To tackle these challenges, we propose a novel hybrid noise strategy, including noises on both dihedral angel and coordinate. However, denoising such hybrid noise in a traditional way is no more equivalent to learning the force field. Through theoretical deductions, we find that the problem is caused by the dependency of the input conformation for covariance. To this end, we propose to decouple the two types of noise and design a novel fractional denoising method (Frad), which only denoises the latter coordinate part. In this way, Frad enjoys both the merits of sampling more low-energy structures and the force field equivalence. Extensive experiments show the effectiveness of Frad in molecule representation, with a new state-of-the-art on 9 out of 12 tasks of QM9 and on 7 out of 8 targets of MD17.

We consider a contextual combinatorial bandit problem where in each round a learning agent selects a subset of arms and receives feedback on the selected arms according to their scores. The score of an arm is an unknown function of the arm's feature. Approximating this unknown score function with deep neural networks, we propose algorithms: Combinatorial Neural UCB ($\texttt{CN-UCB}$) and Combinatorial Neural Thompson Sampling ($\texttt{CN-TS}$). We prove that $\texttt{CN-UCB}$ achieves $\tilde{\mathcal{O}}(\tilde{d} \sqrt{T})$ or $\tilde{\mathcal{O}}(\sqrt{\tilde{d} T K})$ regret, where $\tilde{d}$ is the effective dimension of a neural tangent kernel matrix, $K$ is the size of a subset of arms, and $T$ is the time horizon. For $\texttt{CN-TS}$, we adapt an optimistic sampling technique to ensure the optimism of the sampled combinatorial action, achieving a worst-case (frequentist) regret of $\tilde{\mathcal{O}}(\tilde{d} \sqrt{TK})$. To the best of our knowledge, these are the first combinatorial neural bandit algorithms with regret performance guarantees. In particular, $\texttt{CN-TS}$ is the first Thompson sampling algorithm with the worst-case regret guarantees for the general contextual combinatorial bandit problem. The numerical experiments demonstrate the superior performances of our proposed algorithms.

Symbolic regression (SR) is the problem of learning a symbolic expression from numerical data. Recently, deep neural models trained on procedurally-generated synthetic datasets showed competitive performance compared to more classical Genetic Programming (GP) ones. Unlike their GP counterparts, these neural approaches are trained to generate expressions from datasets given as context. This allows them to produce accurate expressions in a single forward pass at test time. However, they usually do not benefit from search abilities, which result in low performance compared to GP on out-of-distribution datasets. In this paper, we propose a novel method which provides the best of both worlds, based on a Monte-Carlo Tree Search procedure using a context-aware neural mutation model, which is initially pre-trained to learn promising mutations, and further refined from successful experiences in an online fashion. The approach demonstrates state-of-the-art performance on the well-known SRBench benchmark.

In federated learning (FL), a cluster of local clients are chaired under the coordination of the global server and cooperatively train one model with privacy protection. Due to the multiple local updates and the isolated non-iid dataset, clients are prone to overfit into their own optima, which extremely deviates from the global objective and significantly undermines the performance. Most previous works only focus on enhancing the consistency between the local and global objectives to alleviate this prejudicial client drifts from the perspective of the optimization view, whose performance would be prominently deteriorated on the high heterogeneity. In this work, we propose a novel and general algorithm FedSMOO by jointly considering the optimization and generalization targets to efficiently improve the performance in FL. Concretely, FedSMOO adopts a dynamic regularizer to guarantee the local optima towards the global objective, which is meanwhile revised by the global Sharpness Aware Minimization (SAM) optimizer to search for the consistent flat minima. Our theoretical analysis indicates that FedSMOO achieves fast $\mathcal{O}(1/T)$ convergence rate with low generalization bound. Extensive numerical studies are conducted on the real-world dataset to verify its peerless efficiency and excellent generality.

Multi-player online games depict the interaction of multiple players with each other over time. Strongly monotone games are of particular interest since they have benign properties and also relate to many classic games that have applications in real life. Existing works mainly focus on the time-invariant case with provable guarantees established. However, the research of the more general *time-varying* games in changing environments is underexplored and the best-known result cannot match the guarantees in the time-invariant case. In this work, we present a new decentralized online algorithm for time-varying strongly monotone games, which greatly improves existing results and obtains fast rates, matching the best time-invariant guarantee without knowing the environmental non-stationarity. Furthermore, to achieve faster rates, we generalize the RVU property with smoothness and establish a series of problem-dependent bounds that also match the best time-invariant one. To realize all those results, we make a comprehensive use of the techniques in non-stationary and universal online learning.

Parameter inference, i.e. inferring the posterior distribution of the parameters of a statistical model given some data, is a central problem to many scientific disciplines. Posterior inference with generative models is an alternative to methods such as Markov Chain Monte Carlo, both for likelihood-based and simulation-based inference. However, assessing the accuracy of posteriors encoded in generative models is not straightforward. In this paper, we introduce "Tests of Accuracy with Random Points" (TARP) coverage testing as a method to estimate coverage probabilities of generative posterior estimators. Our method differs from previously-existing coverage-based methods, which require posterior evaluations. We prove that our approach is necessary and sufficient to show that a posterior estimator is accurate. We demonstrate the method on a variety of synthetic examples, and show that TARP can be used to test the results of posterior inference analyses in high-dimensional spaces. We also show that our method can detect inaccurate inferences in cases where existing methods fail.

Various optimal gradient-based algorithms have been developed for smooth nonconvex optimization. However, many nonconvex machine learning problems do not belong to the class of smooth functions and therefore the existing algorithms are sub-optimal. Instead, these problems have been shown to satisfy certain generalized-smooth conditions, which have not been well understood in the existing literature. In this paper, we propose a notion of $\alpha$-symmetric generalized-smoothness that substantially extends the existing notions and covers many important functions such as high-order polynomials and exponential functions. We study the fundamental properties and establish descent lemmas for the functions in this class. Then, to solve such a large class of nonconvex problems, we design a special deterministic normalized gradient descent algorithm that achieves the optimal iteration complexity $\mathcal{O}(\epsilon^{-2})$, and also prove that the popular SPIDER variance reduction algorithm achieves the optimal sample complexity $\mathcal{O}(\epsilon^{-3})$. Our results show that solving generalized-smooth nonconvex problems is as efficient as solving smooth nonconvex problems.

This paper studies the challenging continual learning (CL) setting of Class Incremental Learning (CIL). CIL learns a sequence of tasks consisting of disjoint sets of concepts or classes. At any time, a single model is built that can be applied to predict/classify test instances of any classes learned thus far without providing any task related information for each test instance. Although many techniques have been proposed for CIL, they are mostly empirical. It has been shown recently that a strong CIL system needs a strong within-task prediction (WP) and a strong out-of-distribution (OOD) detection for each task. However, it is still not known whether CIL is actually learnable. This paper shows that CIL is learnable. Based on the theory, a new CIL algorithm is also proposed. Experimental results demonstrate its effectiveness.

Horizon dependence is an important difference between reinforcement learning and other machine learning paradigms. Yet, existing results tackling the (exact) horizon dependence either assume that the reward is bounded per step, introducing unfair comparison, or assume strict total boundedness that requires the sum of rewards to be bounded *almost surely* -- allowing only restricted noise on the reward observation. This paper addresses these limitations by introducing a new relaxation -- *expected boundedness* on rewards, where we allow the reward to be stochastic with only boundedness on the *expected* sum -- opening the door to study horizon-dependence with a much broader set of reward functions with noises. We establish a novel generic algorithm that achieves *no-horizon dependence* in terms of sample complexity for both Markov Decision Processes (MDP) and Games, via reduction to a good-conditioned *auxiliary Markovian environment*, in which only ``important'' state-action pairs are preserved. The algorithm takes only $\tilde{O}(\frac{S^2A}{\epsilon^2})$ episodes interacting with such an environment to achieve an $\epsilon$-optimal policy/strategy (with high probability), improving (zhang, 2022) (which only applies to MDPs with deterministic rewards). Here $S$ is the number of states and $A$ is the number of actions, and the bound is independent of the horizon $H$.

Model-based reinforcement learning (RL) often achieves higher sample efficiency in practice than model-free RL by learning a dynamics model to generate samples for policy learning. Previous works learn a dynamics model that fits under the empirical state-action visitation distribution for all historical policies, i.e., the sample replay buffer. However, in this paper, we observe that fitting the dynamics model under the distribution for *all historical policies* does not necessarily benefit model prediction for the *current policy* since the policy in use is constantly evolving over time. The evolving policy during training will cause state-action visitation distribution shifts. We theoretically analyze how this distribution shift over historical policies affects the model learning and model rollouts. We then propose a novel dynamics model learning method, named *Policy-adapted Dynamics Model Learning (PDML)*. PDML dynamically adjusts the historical policy mixture distribution to ensure the learned model can continually adapt to the state-action visitation distribution of the evolving policy. Experiments on a range of continuous control environments in MuJoCo show that PDML achieves significant improvement in sample efficiency and higher asymptotic performance combined with the state-of-the-art model-based RL methods.

We investigate the extent to which offline demonstration data can improve online learning. It is natural to expect some improvement, but *the question is how, and by how much?* We show that the degree of improvement must depend on the *quality* of the demonstration data. To generate portable insights, we focus on Thompson sampling (TS) applied to a multi-armed bandit as a prototypical online learning algorithm and model. The demonstration data is generated by an expert with a given *competence* level, a notion we introduce. We propose an informed TS algorithm that utilizes the demonstration data in a coherent way through Bayes' rule and derive a prior-dependent Bayesian regret bound. This offers insight into how pretraining can greatly improve online performance and how the degree of improvement increases with the expert's competence level. We also develop a practical, approximate informed TS algorithm through Bayesian bootstrapping and show substantial empirical regret reduction through experiments.