Title: Consensus-Driven Active Model Selection
The widespread availability of off-the-shelf machine learning models poses a challenge: which model, of the many available candidates, should be chosen for a given data analysis task? This question of model selection is traditionally answered by collecting and annotating a validation dataset—a costly and time-intensive process. We propose a method for active model selection, using predictions from candidate models to prioritize the labeling of test data points that efficiently differentiate the best candidate. Our method, CODA, performs consensus-driven active model selection by modeling relationships between classifiers, categories, and data points within a probabilistic framework. The framework uses the consensus and disagreement between models in the candidate pool to guide the label acquisition process, and Bayesian inference to update beliefs about which model is best as more information is collected. We validate our approach by curating a collection of 26 benchmark tasks capturing a range of model selection scenarios. CODA outperforms existing methods for active model selection significantly, reducing the annotation effort required to discover the best model by upwards of 70% compared to the previous state-of-the-art. Code and data are available at: https://github.com/justinkay/coda.  more » « less
Award ID(s):
2441060 2330423
PAR ID:
10684798
Author(s) / Creator(s):
 ;  ;  ;  ;  
Publisher / Repository:
IEEE
Date Published:
ISBN:
979-8-3315-8775-8
Page Range / eLocation ID:
4594 to 4604
Subject(s) / Keyword(s):
Location:
Honolulu, HI, USA
Sponsoring Org:
National Science Foundation
More Like this
  1. Integrating regularization methods with standard loss functions such as the least squares, hinge loss, etc., within a regression framework has become a popular choice for researchers to learn predictive models with lower variance and better generalization ability. Regularizers also aid in building interpretable models with high-dimensional data which makes them very appealing. It is observed that each regularizer is uniquely formulated in order to capture data-specific properties such as correlation, structured sparsity and temporal smoothness. The problem of obtaining a consensus among such diverse regularizers while learning a predictive model is extremely important in order to determine the optimal regularizer for the problem. The advantage of such an approach is that it preserves the simplicity of the final model learned by selecting a single candidate model which is not the case with ensemble methods as they use multiple candidate models for prediction. This is called the consensus regularization problem which has not received much attention in the literature due to the inherent difficulty associated with learning and selecting a model from an integrated regularization framework. To solve this problem, in this paper, we propose a method to generate a committee of non-convex regularized linear regression models, and use a consensus criterion to determine the optimal model for prediction. Each corresponding non-convex optimization problem in the committee is solved efficiently using the cyclic-coordinate descent algorithm with the generalized thresholding operator. Our Consensus RegularIzation Selection based Prediction (CRISP) model is evaluated on electronic health records (EHRs) obtained from a large hospital for the congestive heart failure readmission prediction problem. We also evaluate our model on high-dimensional synthetic datasets to assess its performance. The results indicate that CRISP outperforms several state-of-the-art methods such as additive, interactions-based and other competing non-convex regularized linear regression methods. 
    more » « less
  2. Machine-learning (ML) approaches have proven to be of great utility in modern materials innovation pipelines. Generally, ML models are trained on predetermined past data and then used to make predictions for new test cases. Active-learning, however, is a paradigm in which ML models can direct the learning process itself through providing dynamic suggestions/queries for the “next-best experiment.” In this work, the authors demonstrate how an active-learning framework can aid in the discovery of polymers possessing high glass transition temperatures ( T g ). Starting from an initial small dataset of polymer T g measurements, the authors use Gaussian process regression in conjunction with an active-learning framework to iteratively add T g measurements of candidate polymers to the training dataset. The active-learning framework employs one of three decision making strategies (exploitation, exploration, or balanced exploitation/exploration) for selection of the “next-best experiment.” The active-learning workflow terminates once 10 polymers possessing a T g greater than a certain threshold temperature are selected. The authors statistically benchmark the performance of the aforementioned three strategies (against a random selection approach) with respect to the discovery of high- T g polymers for this particular demonstrative materials design challenge. 
    more » « less
  3. While training models and labeling data are resource-intensive, a wealth of pre-trained models and unlabeled data exists. To effectively utilize these resources, we present an approach to actively select pre-trained models while minimizing labeling costs. We frame this as an online contextual active model selection problem: At each round, the learner receives an unlabeled data point as a context. The objective is to adaptively select the best model to make a prediction while limiting label requests. To tackle this problem, we propose CAMS, a contextual active model selection algorithm that relies on two novel components: (1) a contextual model selection mechanism, which leverages context information to make informed decisions about which model is likely to perform best for a given context, and (2) an active query component, which strategically chooses when to request labels for data points, minimizing the overall labeling cost. We provide rigorous theoretical analysis for the regret and query complexity under both adversarial and stochastic settings. Furthermore, we demonstrate the effectiveness of our algorithm on a diverse collection of benchmark classification tasks. Notably, CAMS requires substantially less labeling effort (less than 10%) compared to existing methods on CIFAR10 and DRIFT benchmarks, while achieving similar or better accuracy. 
    more » « less
  4. In this thesis, I present a decentralized sparse Gaussian process regression (DSGPR) model with event-triggered, adaptive inducing points. This DSGPR model brings the advantages of sparse Gaussian process regression to a decentralized implementation. Being decentralized and sparse provides advantages that are ideal for multi-agent systems (MASs) performing environmental modeling. In this case, MASs need to model large amounts of information while having potential intermittent communication connections. Additionally, the model needs to correctly perform uncertainty propagation between autonomous agents and ensure high accuracy on the prediction. For the model to meet these requirements, a bounded and efficient real-time sparse Gaussian process regression (SGPR) model is needed. I improve real-time SGPR models in these regards by introducing an adaptation of the mean shift and fixed-width clustering algorithms called radial clustering. Radial clustering enables real-time SGPR models to have an adaptive number of inducing points through an efficient inducing point selection process. I show how this clustering approach scales better than other seminal Gaussian process regression (GPR) and SGPR models for real-time purposes while attaining similar prediction accuracy and uncertainty reduction performance. Furthermore, this thesis addresses common issues inherent in decentralized frameworks such as high computation costs, inter-agent message bandwidth restrictions, and data fusion integrity. These challenges are addressed in part through performing maximum consensus between local agent models which enables the MAS to gain the advantages of decentralization while keeping data fusion integrity. The inter-agent communication restrictions are addressed through the contribution of two message passing heuristics called the covariance reduction heuristic and the Bhattacharyya distance heuristic. These heuristics enable user to reduce message passing frequency and message size through the Bhattacharyya distance and properties of spatial kernels. The entire DSGPR framework is evaluated on multiple simulated random vector fields. The results show that this framework effectively estimates vector fields using multiple autonomous agents. This vector field is assumed to be a wind field; however, this framework may be applied to the estimation of other scalar or vector fields (e.g., fluids, magnetic fields, electricity, etc.). Keywords: Sparse Gaussian process regression, clustering, event-triggered, decentralized, sensor fusion, uncertainty propagation, inducing points 
    more » « less
  5. Summary Model selection is crucial both to high-dimensional learning and to inference for contemporary big data applications in pinpointing the best set of covariates among a sequence of candidate interpretable models. Most existing work implicitly assumes that the models are correctly specified or have fixed dimensionality, yet both model misspecification and high dimensionality are prevalent in practice. In this paper, we exploit the framework of model selection principles under the misspecified generalized linear models presented in Lv & Liu (2014), and investigate the asymptotic expansion of the posterior model probability in the setting of high-dimensional misspecified models. With a natural choice of prior probabilities that encourages interpretability and incorporates the Kullback–Leibler divergence, we suggest using the high-dimensional generalized Bayesian information criterion with prior probability for large-scale model selection with misspecification. Our new information criterion characterizes the impacts of both model misspecification and high dimensionality on model selection. We further establish the consistency of covariance contrast matrix estimation and the model selection consistency of the new information criterion in ultrahigh dimensions under some mild regularity conditions. Our numerical studies demonstrate that the proposed method enjoys improved model selection consistency over its main competitors. 
    more » « less