skip to main content

Explore Research Products in the NSF-PAR

Title: Contextual Active Online Model Selection with Expert Advice
How can we collect the most useful labels to learn a model selection policy, when presented with arbitrary heterogeneous data streams? In this paper, we formulate this task as a contextual active model selection problem, where at each round the learner receives an unlabeled data point along with a context. The goal is to output the best model for any given context without obtaining an excessive amount of labels. In particular, we focus on the task of selecting pre-trained classifiers, and propose a contextual active model selection algorithm (CAMS), which relies on a novel uncertainty sampling query criterion defined on a given policy class for adaptive model selection. In comparison to prior art, our algorithm does not assume a globally optimal model. We provide rigorous theoretical analysis for the regret and query complexity under both adversarial and stochastic settings. Our experiments on several benchmark classification datasets demonstrate the algorithm’s effectiveness in terms of both regret and query complexity. Notably, to achieve the same accuracy, CAMS incurs less than 10% of the label cost when compared to the best online model selection baselines on CIFAR10.  more » « less
Award ID(s):
2041970
NSF-PAR ID:
10487005
Author(s) / Creator(s):
; ; ;
Publisher / Repository:
ICML
Date Published:
Journal Name:
ICML2022 Workshop on Adaptive Experimental Design and Active Learning in the Real World
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ( , Advances in neural information processing systems)
    Koyejo, S. ; Mohamed, S. ; Agarwal, A. ; Belgrave, D. ; Cho, K. ; Oh, A. (Ed.)
    In the stochastic contextual bandit setting, regret-minimizing algorithms have been extensively researched, but their instance-minimizing best-arm identification counterparts remain seldom studied. In this work, we focus on the stochastic bandit problem in the (ǫ, δ)-PAC setting: given a policy class Π the goal of the learner is to return a policy π ∈ Π whose expected reward is within ǫ of the optimal policy with probability greater than 1 − δ. We characterize the first instance-dependent PAC sample complexity of contextual bandits through a quantity ρΠ, and provide matching upper and lower bounds in terms of ρΠ for the agnostic and linear contextual best-arm identification settings. We show that no algorithm can be simultaneously minimax-optimal for regret minimization and instance-dependent PAC for best-arm identification. Our main result is a new instance-optimal and computationally efficient algorithm that relies on a polynomial number of calls to an argmax oracle. 
    more » « less
  2. ( , ICML)
    We consider the classical problem of multi-class prediction with expert advice, but with an active learning twist. In this new setting the learner will only query the labels of a small number of examples, but still aims to minimize regret to the best expert as usual; the learner is also allowed a very short "burn-in" phase where it can fast-forward and query certain highly-informative examples. We design an algorithm that utilizes Hedge (aka Exponential Weights) as a subroutine, and we show that under a very particular combinatorial constraint on the matrix of expert predictions we can obtain a very strong regret guarantee while querying very few labels. This constraint, which we refer to as ζ -compactness, or just compactness, can be viewed as a non-stochastic variant of the disagreement coefficient, another popular parameter used to reason about the sample complexity of active learning in the IID setting. We also give a polynomial-time algorithm to calculate the ζ -compactness of a matrix up to an approximation factor of 3. 
    more » « less
  3. ; ; ( , SPAA '20: Proceedings of the 32nd ACM Symposium on Parallelism in Algorithms and Architectures)
    The K-nearest neighbors is a basic problem in machine learning with numerous applications. In this problem, given a (training) set of n data points with labels and a query point q, we want to assign a label to q based on the labels of the K-nearest points to the query. We study this problem in the k-machine model, a model for distributed large-scale data. In this model, we assume that the n points are distributed (in a balanced fashion) among the k machines and the goal is to compute an answer given a query point to a machine using a small number of communication rounds. Our main result is a randomized algorithm in the k-machine model that runs in O(log K) communication rounds with high success probability (regardless of the number of machines k and the number of points n). The message complexity of the algorithm is small taking only O(k log K) messages. Our bounds are essentially the best possible for comparison-based algorithms. We also implemented our algorithm and show that it performs well in practice. 
    more » « less
  4. ; ; ( , Management Science)
    In this paper, we study the learning problem in contextual search, which is motivated by applications such as crowdsourcing and personalized medicine experiments. In particular, for a sequence of arriving context vectors, with each context associated with an underlying value, the decision maker either makes a query at a certain point or skips the context. The decision maker will only observe the binary feedback on the relationship between the query point and the value associated with the context. We study a probably approximately correct learning setting, where the goal is to learn the underlying mean value function in context with a minimum number of queries. To address this challenge, we propose a trisection search approach combined with a margin-based active learning method. We show that the algorithm only needs to make [Formula: see text] queries to achieve an ε-estimation accuracy. This sample complexity significantly reduces the required sample complexity in the passive setting where neither sample skipping nor query selection is allowed, which is at least [Formula: see text]. This paper was accepted by J. George Shanthikumar, data science. 
    more » « less
  5. ; ( , Electronic journal of statistics)
    In this paper, the problem of maximizing a black-box function f:→ℝ is studied in the Bayesian framework with a Gaussian Process prior. In particular, a new algorithm for this problem is proposed, and high probability bounds on its simple and cumulative regret are established. The query point selection rule in most existing methods involves an exhaustive search over an increasingly fine sequence of uniform discretizations of . The proposed algorithm, in contrast, adaptively refines  which leads to a lower computational complexity, particularly when  is a subset of a high dimensional Euclidean space. In addition to the computational gains, sufficient conditions are identified under which the regret bounds of the new algorithm improve upon the known results. Finally, an extension of the algorithm to the case of contextual bandits is proposed, and high probability bounds on the contextual regret are presented. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email