Sublinear Partition Estimation Article Swipe

PDF

Related Concepts

Discriminative model Sublinear function Categorical variable Pattern recognition (psychology) Partition (number theory) Computer science Artificial intelligence Classifier (UML) Artificial neural network Mathematics Algorithm Machine learning Discrete mathematics Combinatorics

Pushpendre Rastogi , Benjamin Van Durme ·

YOU? · · 2015 · Open Access · · DOI: https://doi.org/10.48550/arxiv.1508.01596 · OA: W1963476456

The output scores of a neural network classifier are converted to probabilities via normalizing over the scores of all competing categories. Computing this partition function, $Z$, is then linear in the number of categories, which is problematic as real-world problem sets continue to grow in categorical types, such as in visual object recognition or discriminative language modeling. We propose three approaches for sublinear estimation of the partition function, based on approximate nearest neighbor search and kernel feature maps and compare the performance of the proposed approaches empirically.

Keywords: Discriminative model · Sublinear function · Categorical variable · Pattern recognition (psychology) · Partition (number theory) · Computer science · Artificial intelligence · Classifier (UML) · Artificial neural network · Mathematics · Algorithm · Machine learning · Discrete mathematics · Combinatorics