Convolutional Tables Ensemble: classification in microseconds

TL;DR

This paper introduces Convolutional Tables Ensemble (CTE), a fast classification architecture optimized for microsecond-level inference, achieving high accuracy and speed advantages over CNNs on object recognition tasks.

Contribution

The paper presents a novel fast classification architecture, CTE, with new decision functions, a tree learning algorithm, and CNN distillation, enabling ultra-fast object recognition.

Findings

Achieves 24-45% accuracy improvements over similar-speed methods.

Outperforms CNNs in speed-accuracy trade-offs within certain time constraints.

Provides 5-200X speedup with comparable error rates to CNNs.

Abstract

We study classifiers operating under severe classification time constraints, corresponding to 1-1000 CPU microseconds, using Convolutional Tables Ensemble (CTE), an inherently fast architecture for object category recognition. The architecture is based on convolutionally-applied sparse feature extraction, using trees or ferns, and a linear voting layer. Several structure and optimization variants are considered, including novel decision functions, tree learning algorithm, and distillation from CNN to CTE architecture. Accuracy improvements of 24-45% over related art of similar speed are demonstrated on standard object recognition benchmarks. Using Pareto speed-accuracy curves, we show that CTE can provide better accuracy than Convolutional Neural Networks (CNN) for a certain range of classification time constraints, or alternatively provide similar error rates with 5-200X speedup.