SpeedMachines: Anytime Structured Prediction
This addresses the need for efficient, anytime structured prediction in applications like computer vision, where computational constraints are critical, though it appears incremental as it builds on existing structured prediction methods.
The paper tackles the problem of making structured prediction models computationally efficient for anytime inference, where predictions must be made within a budget and improve over time, by proposing a technique that incorporates structural and feature computation trade-offs during training. It demonstrates this on scene understanding in computer vision, showing predictions that gradually approach state-of-the-art classification performance as time increases.
Structured prediction plays a central role in machine learning applications from computational biology to computer vision. These models require significantly more computation than unstructured models, and, in many applications, algorithms may need to make predictions within a computational budget or in an anytime fashion. In this work we propose an anytime technique for learning structured prediction that, at training time, incorporates both structural elements and feature computation trade-offs that affect test-time inference. We apply our technique to the challenging problem of scene understanding in computer vision and demonstrate efficient and anytime predictions that gradually improve towards state-of-the-art classification performance as the allotted time increases.