Stochastic Maximum Likelihood Optimization via Hypernetworks
This addresses optimization challenges in neural networks for general ML tasks, but it is incremental as it builds on existing hypernetwork methods.
The paper tackles maximum likelihood optimization of neural networks by using hypernetworks to initialize weights, achieving competitive results on regression and classification benchmarks.
This work explores maximum likelihood optimization of neural networks through hypernetworks. A hypernetwork initializes the weights of another network, which in turn can be employed for typical functional tasks such as regression and classification. We optimize hypernetworks to directly maximize the conditional likelihood of target variables given input. Using this approach we obtain competitive empirical results on regression and classification benchmarks.