We establish a new theoretical framework for learning under multi-class, instancedependent label noise. This framework casts learning with label noise as a form of domain adaptation, in particular, domain adaptation under posterior drift. We introduce the concept of relative signal strength (RSS), a pointwise measure that quantifies the transferability from noisy to clean posterior. Using RSS, we establish nearly matching upper and lower bounds on the excess risk. Our theoretical findings support the simple Noise Ignorant Empirical Risk Minimization (NI-ERM) principle, which minimizes empirical risk while ignoring label noise. Finally, we translate this theoretical insight into practice: by using NI-ERM to fit a linear classifier on top of a self-supervised feature extractor, we achieve state-of-the-art performance on the CIFAR-N data challenge.

Recently, differentiable mask pruning methods optimize the continuous relaxation architecture (soft network) as the proxy of the pruned discrete network (hard network) for superior sub-architecture search. However, due to the agnostic impact of the discretization process, the hard network struggles with the equivalent representational capacity as the soft network, namely discretization gap, which severely spoils the pruning performance. In this paper, we first investigate the discretization gap and propose a novel structural differentiable mask pruning framework named S2HPruner to bridge the discretization gap in a one-stage manner.

We consider a variant of the multiarmed bandit problem where jobs queue for service, and service rates of different servers may be unknown. We study algorithms that minimize queue-regret: the (expected) difference between the queue-lengths obtained by the algorithm, and those obtained by a "genie"-aided matching algorithm that knows exact service rates. A naive view of this problem would suggest that queue-regret should grow logarithmically: since queue-regret cannot be larger than classical regret, results for the standard MAB problem give algorithms that ensure queue-regret increases no more than logarithmically in time. Our paper shows surprisingly more complex behavior. In particular, the naive intuition is correct as long as the bandit algorithm's queues have relatively long regenerative cycles: in this case queue-regret is similar to cumulative regret, and scales (essentially) logarithmically. However, we show that this "early stage" of the queueing bandit eventually gives way to a "late stage", where the optimal queue-regret scaling is O(1/t). We demonstrate an algorithm that (order-wise) achieves this asymptotic queue-regret, and also exhibits close to optimal switching time from the early stage to the late stage.

Using RL for improving the performance of systems has a lot of potential, but is also in many ways very different from, for example, using RL for games. Thus, in this work we first discuss the unique challenges RL for systems has, and then propose Park an open extensible platform, which makes it easier for ML researchers to work on systems problems. Currently, Park consists of 12 real world system-centric optimization problems with one common easy to use interface. Finally, we present the performance of existing RL approaches over those 12 problems and outline potential areas of future work.

Thank you for raising this point.

The growing popularity of transfer learning, due to the availability of models pre-trained on vast amounts of data, makes it imperative to understand when the knowledge of these pre-trained models can be transferred to obtain high-performing models on downstream target tasks. However, the exact conditions under which transfer learning succeeds in a cross-domain cross-task setting are still poorly understood. To bridge this gap, we propose a novel analysis that analyzes the transferability of the representations of pre-trained models to downstream tasks in terms of their relatedness to a given reference task. Our analysis leads to an upper bound on transferability in terms of task-relatedness, quantified using the difference between the class priors, label sets, and features of the two tasks. Our experiments using state-of-the-art pre-trained models show the effectiveness of task-relatedness in explaining transferability on various vision and language tasks. The efficient computability of task-relatedness even without labels of the target task and its high correlation with the model's accuracy after end-to-end fine-tuning on the target task makes it a useful metric for transferability estimation.

To reduce annotation costs, it is common in crowdsourcing to collect only a few noisy labels from different crowd workers for each instance. However, the limited noisy labels restrict the performance of label integration algorithms in inferring the unknown true label for the instance. Recent works have shown that leveraging neighbor instances can help alleviate this problem. Yet, these works all assume that each instance has the same neighborhood size, which defies common sense. To address this gap, we propose a novel label integration algorithm called K-free nearest neighbor (KFNN). In KFNN, the neighborhood size of each instance is automatically determined based on its attributes and noisy labels.

We study the stochastic optimization of canonical correlation analysis (CCA), whose objective is nonconvex and does not decouple over training samples. Although several stochastic gradient based optimization algorithms have been recently proposed to solve this problem, no global convergence guarantee was provided by any of them. Inspired by the alternating least squares/power iterations formulation of CCA, and the shift-and-invert preconditioning method for PCA, we propose two globally convergent meta-algorithms for CCA, both of which transform the original problem into sequences of least squares problems that need only be solved approximately. We instantiate the meta-algorithms with state-of-the-art SGD methods and obtain time complexities that significantly improve upon that of previous work. Experimental results demonstrate their superior performance.

We propose a new complexity measure for Markov decision processes (MDPs), the maximum expected hitting cost (MEHC). This measure tightens the closely related notion of diameter [JOA10] by accounting for the reward structure. We show that this parameter replaces diameter in the upper bound on the optimal value span of an extended MDP, thus refining the associated upper bounds on the regret of several UCRL2-like algorithms. Furthermore, we show that potential-based reward shaping [NHR99] can induce equivalent reward functions with varying informativeness, as measured by MEHC. We further establish that shaping can reduce or increase MEHC by at most a factor of two in a large class of MDPs with finite MEHC and unsaturated optimal average rewards.

We thank the reviewers for their thoughtful comments and helpful suggestions, and respond to comments below. R1: "Interesting theoretical connections of maximum expected hitting This work may be useful in practice to craft informative shaping rewards" R2: "The paper is well-written and provides an interesting insight on the fact that the R1: "bugs" in designed rewards (minor comment 1) - We appreciate your suggestion and we shall add an explicit reference in any future versions. In contrast, in prior works, PBRS's learning We are also curious about the same question for non-optimism-based algorithms. We plan to pursue it in future works. This is an interesting suggestion and we will consider it seriously.