Timezone: »

Feature Learning and Signal Propagation in Deep Neural Networks
Yizhang Lou · Chris Mingard · Soufiane Hayou

Wed Jul 20 08:00 AM -- 08:05 AM (PDT) @ Room 309

Recent work by Baratin et al. (2021) sheds light on an intriguing pattern that occurs during the training of deep neural networks: some layers align much more with data compared to other layers (where the alignment is defined as the normalize euclidean product of the tangent features matrix and the data labels matrix). The curve of the alignment as a function of layer index (generally) exhibits a ascent-descent pattern where the maximum is reached for some hidden layer. In this work, we provide the first explanation for this phenomenon. We introduce the Equilibrium Hypothesis which connects this alignment pattern to signal propagation in deep neural networks. Our experiments demonstrate an excellent match with the theoretical predictions.

Author Information

Yizhang Lou (University of Oxford)
Chris Mingard (University of Oxford)
Soufiane Hayou (National University of Singapore)

Related Events (a corresponding poster, oral, or spotlight)