Date of Award




Document Type


Degree Name

Doctor of Philosophy (PhD)


Department of Electrical and Computer Engineering

Content Description

1 online resource (xvi, 124 pages) : color illustrations.

Dissertation/Thesis Chair

Daphney-Stavroula Zois

Committee Members

Charalampos Chelmis, Aveek Dutta, Gary Saulnier


Machine learning, Artificial intelligence, Decision making, Ensemble learning (Machine learning)

Subject Categories

Computer Engineering


In a typical supervised machine learning setting, the predictions on all test instances are based on a common subset of features discovered during model training. However, using a different subset of features that are most informative for each test instance individually may improve not only the quality of prediction but also the overall interpretability of the model. To this end, in this dissertation, we study the problem of optimizing the trade-off between instance-level sparsity and the quality of prediction using a dynamic instance-wise decision-making approach. Specifically, this approach sequentially reviews features one at a time for each data instance given previously chosen features and stops this process to predict the instance once it determines that including additional features will not improve the quality of final decision. In contrast to most existing work that utilizes a set of features common for all data instances, this method utilizes different features to predict different data instances.