Combinets: Creativity via Recombination of Neural Networks
This addresses the issue of handling cases outside training data for problems with limited data, which is incremental as it builds on existing transfer learning methods.
The paper tackles the problem of deep neural networks struggling with limited training data by introducing conceptual expansion, a general representation for reusing trained models to derive new ones without backpropagation. It outperforms standard transfer learning approaches on few-shot image classification and generation tasks.
One of the defining characteristics of human creativity is the ability to make conceptual leaps, creating something surprising from typical knowledge. In comparison, deep neural networks often struggle to handle cases outside of their training data, which is especially problematic for problems with limited training data. Approaches exist to transfer knowledge from problems with sufficient data to those with insufficient data, but they tend to require additional training or a domain-specific method of transfer. We present a new approach, conceptual expansion, that serves as a general representation for reusing existing trained models to derive new models without backpropagation. We evaluate our approach on few-shot variations of two tasks: image classification and image generation, and outperform standard transfer learning approaches.