Stochastic Maximum Likelihood Optimization via Hypernetworks

2017/12/04 by Abdul-Saboor Sheikh, Kashif Rasul, Sheikh, Abdul-Saboor +5
Computer Science · #FOS: Computer and information sciences #Gaussian Processes and Bayesian Inference #Generative Adversarial Networks and Image Synthesis #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Neural Networks and Applications

paper · pdf · doi:10.48550/arxiv.1712.01141

openalex publication_date 2017/12/04 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

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.

Citations

• Stein Variational Gradient Descent: A General Purpose Bayesian Inference\n Algorithm
• Dropout Inference in Bayesian Neural Networks with Alpha-divergences
• Weight Uncertainty in Neural Networks
• Adversarial Variational Bayes: Unifying Variational Autoencoders and Generative Adversarial Networks

Related

• Stochastic Hyperparameter Optimization through Hypernetworks
• Principled Weight Initialization for Hypernetworks
• Hyperparameter Optimization through Neural Network Partitioning
• Bayesian Hypernetworks
• Hyper-Parameter Optimization: A Review of Algorithms and Applications