IDK Cascades: Fast Deep Learning by Learning not to Overthink

Abstract

Advances in deep learning have led to substantial increases in prediction accuracy but have been accompanied by increases in the cost of rendering predictions. We conjecture that fora majority of real-world inputs, the recent advances in deep learning have created models that effectively "overthink" on simple inputs. In this paper, we revisit the classic question of building model cascades that primarily leverage class asymmetry to reduce cost. We introduce the "I Don't Know"(IDK) prediction cascades framework, a general framework to systematically compose a set of pre-trained models to accelerate inference without a loss in prediction accuracy. We propose two search based methods for constructing cascades as well as a new cost-aware objective within this framework. The proposed IDK cascade framework can be easily adopted in the existing model serving systems without additional model re-training. We evaluate the proposed techniques on a range of benchmarks to demonstrate the effectiveness of the proposed framework.

UAI 2018 camera ready

References in corpus (6)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• TensorFlow-Serving: Flexible, High-Performance ML Serving
• Structured Prediction Cascades
• NoScope: Optimizing Neural Network Queries over Video at Scale
• Deep Layer Aggregation

Cited by in corpus (12)

• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• Adaptive Inference through Early-Exit Networks: Design, Challenges and Directions
• Shallow-Deep Networks: Understanding and Mitigating Network Overthinking
• Model Slicing for Supporting Complex Analytics with Elastic Inference Cost and Resource Constraints
• TAFE-Net: Task-Aware Feature Embeddings for Low Shot Learning
• Fast Real-time Counterfactual Explanations
• SCOUT: Self-aware Discriminant Counterfactual Explanations
• Improving Anytime Prediction with Parallel Cascaded Networks and a Temporal-Difference Loss
• MultiTASC++: A Continuously Adaptive Scheduler for Edge-Based Multi-Device Cascade Inference
• Learning to Cascade: Confidence Calibration for Improving the Accuracy and Computational Cost of Cascade Inference Systems
• When in Doubt, Summon the Titans: Efficient Inference with Large Models
• Pay Attention to Convolution Filters: Towards Fast and Accurate Fine-Grained Transfer Learning