Classification and Reconstruction Processes in Deep Predictive Coding Networks: Antagonists or Allies?
This work addresses a fundamental challenge in predictive coding-inspired deep networks for visual computing, revealing an antagonistic relationship between classification and reconstruction that is incremental but crucial for guiding future model designs.
The study investigated the interaction between classification and reconstruction processes in deep predictive coding networks for visual computing, finding that these processes compete for shared latent representations, with classification-driven information diminishing reconstruction-driven information and vice versa, though increasing representation dimensions or network complexity can mitigate this trade-off.
Predictive coding-inspired deep networks for visual computing integrate classification and reconstruction processes in shared intermediate layers. Although synergy between these processes is commonly assumed, it has yet to be convincingly demonstrated. In this study, we take a critical look at how classifying and reconstructing interact in deep learning architectures. Our approach utilizes a purposefully designed family of model architectures reminiscent of autoencoders, each equipped with an encoder, a decoder, and a classification head featuring varying modules and complexities. We meticulously analyze the extent to which classification- and reconstruction-driven information can seamlessly coexist within the shared latent layer of the model architectures. Our findings underscore a significant challenge: Classification-driven information diminishes reconstruction-driven information in intermediate layers' shared representations and vice versa. While expanding the shared representation's dimensions or increasing the network's complexity can alleviate this trade-off effect, our results challenge prevailing assumptions in predictive coding and offer guidance for future iterations of predictive coding concepts in deep networks.