Federated Topic Model uses Variational Autoencoder and Pruning for faster training

Researchers have developed a novel federated topic modeling approach using variational autoencoders to address privacy concerns in cross-party document analysis. This method incorporates neural network model pruning to accelerate training and inference times. Two pruning strategies are proposed: one that gradually prunes throughout training for higher accuracy and reduced inference time, and another that quickly reaches a target pruning rate early in training for faster completion, potentially with some information loss. Experiments demonstrate that this combined approach significantly speeds up training while preserving model performance. AI

IMPACT This research offers a method to accelerate training and inference for topic models in privacy-sensitive federated learning scenarios.

RANK_REASON The cluster contains a research paper detailing a new method for federated topic modeling. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Ang Li
• arXiv
• Federated Topic Model
• Model Pruning
• variational auto-encoder

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →