• Publications
  • Influence
Adaptive Neural Networks for Efficient Inference
TLDR
It is shown that computational time can be dramatically reduced by exploiting the fact that many examples can be correctly classified using relatively efficient networks and that complex, computationally costly networks are only necessary for a small fraction of examples. Expand
Pruning Random Forests for Prediction on a Budget
TLDR
This work poses pruning RFs as a novel 0-1 integer program with linear constraints that encourages feature re-use and establishes total unimodularity of the constraint set to prove that the corresponding LP relaxation solves the original integer program. Expand
Feature-Budgeted Random Forest
TLDR
A novel random forest algorithm to minimize prediction error for a user-specified feature acquisition budget and demonstrate superior accuracy-cost curves against state-of-the-art prediction-time algorithms. Expand
Local Supervised Learning through Space Partitioning
TLDR
An empirical risk minimization problem that incorporates both partitioning and classification in to a single global objective is formulated and it is shown that space partitioning can be equivalently reformulated as a supervised learning problem and consequently any discriminative learning method can be utilized in conjunction with this approach. Expand
Efficient Learning by Directed Acyclic Graph For Resource Constrained Prediction
TLDR
This work models the problem of reducing test-time acquisition costs in classification systems as a directed acyclic graph (DAG) where internal nodes correspond to sensor subsets and decision functions at each node choose whether to acquire a new sensor or classify using the available measurements. Expand
Adaptive Neural Networks for Fast Test-Time Prediction
TLDR
An adaptive network evaluation scheme is posed, where a system to adaptively choose the components of a deep network to be evaluated for each example, to reduce the evaluation time on new examples without loss of classification performance. Expand
Energy-Efficient Adaptive Classifier Design for Mobile Systems
TLDR
An adaptive classifier that leverages the wide variability in data complexity to enable energy-efficient data classification operations for mobile systems and takes advantage of varying classification "hardness" across data to dynamically allocate resources and improve energy efficiency. Expand
An LP for Sequential Learning Under Budgets
TLDR
The authors' LP achieves or exceeds the empirical performance of the nonconvex alternating algorithm that requires a large number of random initializations and has the advantage of guaranteed convergence, global optimality, repeatability and computation eciency. Expand
Fast margin-based cost-sensitive classification
TLDR
A novel classification algorithm for learning with test time budgets that provides an accurate estimate of classification confidence and outperforms other approaches while being significantly more efficient in computation. Expand
Locally-Linear Learning Machines (L3M)
TLDR
Improvement in classification performance, test and training time relative to common discriminative learning methods on challenging multiclass data sets is demonstrated. Expand
...
1
2
3
...