The paper in my opinion studies an interesting and relevant problem - one of modelling the tradeoff between information, cost and reward (whether to choose low information that is cheap or high information that is expensive) - in online learning, specifically stochastic bandits. In this sense it may be useful as a benchmark to improve upon. Though the paper seems technically solid, a key shortcoming is the lack of adequate explanation about its results and assumptions. The regret definition adopted seems unnatural at least from one angle - why not penalize resource consumption (or'cost') additively instead of multiplicatively as done here? The authors' example of ad-display motivates their definition, but may not be the most general.

The present work takes a significant step in addressing this. The primary contribution of the paper are variations of Online Newton Step that remove this drawback using a sketching approximation to the scaling matrix and a clever implementation of sparse updates. The primary theoretical contributions are the analysis of the RP and FD versions of the algorithm. For RP they show a regret bound which holds when the matrix G_T (the matrix of observed gradients) is actually low-rank. Given the structure of the loss functions assumed, f_t(w) \ell( w, x_t), gradients will always be in the direction of the examples x_t, and so I think this theorem only holds when the data is actually low-rank.

Online convex minimization is a fundamental problem in learning theory with various applications. Here, we are given a sequence of convex functions f1,...,fT in an online fashion and we seek to predict the minimizer of the sum so as to minimize the regret. This is a very well studied question with various guarantees known even when only given gradient information about the functions fi. For general functions this leads to a regret of sqrt(T) and for some special cases, we know better guarantees: O(log T) for strongly convex functions, O(d log T) for exp-convex functions etc. However, the algorithms that achieve the latter better guarantees are quite different and in some cases need to know bounds on the specific parameters of the functions (e.g., strong-convexity of the fi's, although this can now be avoided).

In many practical machine learning problems, the acquisition of labeled data is often expensive and/or time consuming. This motivates us to study a problem as follows: given a label budget, how to select data points to label such that the learning performance is optimized. We propose a selective labeling method by analyzing the generalization error of Laplacian regularized Least Squares (LapRLS). In particular, we derive a deterministic generalization error bound for LapRLS trained on subsampled data, and propose to select a subset of data points to label by minimizing this upper bound. Since the minimization is a combinational problem, we relax it into continuous domain and solve it by projected gradient descent.

Many practitioners of reinforcement learning problems have observed that oftentimes the performance of the agent reaches very close to the optimal performance even though the estimated (action-)value function is still far from the optimal one. The goal of this paper is to explain and formalize this phenomenon by introducing the concept of the action-gap regularity. As a typical result, we prove that for an agent following the greedy policy (\hat{\pi}) with respect to an action-value function (\hat{Q}), the performance loss (E[V (X) - V {\hat{X}} (X)]) is upper bounded by (O( \hat{Q} - Q _\infty {1 \zeta})), in which (\zeta 0) is the parameter quantifying the action-gap regularity. For (\zeta 0), our results indicate smaller performance loss compared to what previous analyses had suggested. Finally, we show how this regularity affects the performance of the family of approximate value iteration algorithms.

Sign language translation (SLT) aims to convert a continuous sign language video clip into a spoken language. Considering different geographic regions generally have their own native sign languages, it is valuable to establish corresponding SLT datasets to support related communication and research. Auslan, as a sign language specific to Australia, still lacks a dedicated large-scale dataset for SLT.To fill this gap, we curate an Australian Sign Language translation dataset, dubbed Auslan-Daily, which is collected from the Auslan educational TV series and Auslan TV programs. The former involves daily communications among multiple signers in the wild, while the latter comprises sign language videos for up-to-date news, weather forecasts, and documentaries. In particular, Auslan-Daily has two main features: (1) the topics are diverse and signed by multiple signers, and (2) the scenes in our dataset are more complex, e.g., captured in various environments, gesture interference during multi-signers' interactions and various camera positions.

Algorithmic decision-making in high-stakes domains often involves assigning decisions to agents with incentives to strategically modify their input to the algorithm. We formalize this setting as an online learning problem with apple-tasting feedback where a principal makes decisions about a sequence of T agents, each of which is represented by a context that may be strategically modified. Our goal is to achieve sublinear strategic regret, which compares the performance of the principal to that of the best fixed policy in hindsight, if the agents were truthful when revealing their contexts. Our main result is a learning algorithm which incurs \tilde{\mathcal{O}}(\sqrt{T}) strategic regret when the sequence of agents is chosen stochastically. We also give an algorithm capable of handling adversarially-chosen agents, albeit at the cost of \tilde{\mathcal{O}}(T {(d 1)/(d 2)}) strategic regret (where d is the dimension of the context).

Delayed feedback is a critical problem in dynamic recommender systems. In practice, the feedback result often depends on the frequency of recommendation. Most existing online learning literature fails to consider optimization of the recommendation frequency, and regards the reward from each successfully recommended message to be equal. In this paper, we consider a novel cascading bandits setting, where individual messages from a selected list are sent to a user periodically. Whenever a user does not like a message, she may abandon the system with a probability positively correlated with the recommendation frequency.

In this work we revisit an interactive variant of joint differential privacy, recently introduced by Naor et al. [2023], and generalize it towards handling online processes in which existing privacy definitions seem too restrictive. We study basic properties of this definition and demonstrate that it satisfies (suitable variants) of group privacy, composition, and post processing.In order to demonstrate the advantages of this privacy definition compared to traditional forms of differential privacy,we consider the basic setting of online classification. We show that any (possibly non-private) learning rule can be effectively transformed to a private learning rule with only a polynomial overhead in the mistake bound. This demonstrates a stark difference with traditional forms of differential privacy, such as the one studied by Golowich and Livni [2021], where only a double exponential overhead in the mistake bound is known (via an information theoretic upper bound).

Sign languages are used as a primary language by approximately 70 million D/deaf people world-wide. However, most communication technologies operate in spoken and written languages, creating inequities in access. To help tackle this problem, we release ASL Citizen, the first crowdsourced Isolated Sign Language Recognition (ISLR) dataset, collected with consent and containing 83,399 videos for 2,731 distinct signs filmed by 52 signers in a variety of environments. We propose that this dataset be used for sign language dictionary retrieval for American Sign Language (ASL), where a user demonstrates a sign to their webcam to retrieve matching signs from a dictionary. We show that training supervised machine learning classifiers with our dataset advances the state-of-the-art on metrics relevant for dictionary retrieval, achieving 63\% accuracy and a recall-at-10 of 91\%, evaluated entirely on videos of users who are not present in the training or validation sets.