Everything old is new again: A multi-view learning approach to learning using privileged information and distillation
This work provides a theoretical analysis for knowledge transfer methods, which is incremental as it reframes existing settings rather than introducing a new paradigm.
The paper tackles the problem of unifying learning using privileged information (LUPI) and distillation under a multi-view framework, showing that encouraging teacher-student agreement reduces search space and leads to improved convergence rates in regularized empirical risk minimization.
We adopt a multi-view approach for analyzing two knowledge transfer settings---learning using privileged information (LUPI) and distillation---in a common framework. Under reasonable assumptions about the complexities of hypothesis spaces, and being optimistic about the expected loss achievable by the student (in distillation) and a transformed teacher predictor (in LUPI), we show that encouraging agreement between the teacher and the student leads to reduced search space. As a result, improved convergence rate can be obtained with regularized empirical risk minimization.