carpedm20/variational-text-tensorflow

Bad performance in document classification

Open

#2 opened on Jan 12, 2017

View on GitHub
 (1 comment) (0 reactions) (0 assignees)Python (76 forks)github user discovery
enhancementhelp wanted

Repository metrics

Stars
 (535 stars)
PR merge metrics
 (PR metrics pending)

Description

Hi Taehoon,

Thank you for sharing your implementation. Have you tried applying the NVDM model in some downstream tasks like document classification? I did not find such experiments in the original paper, so I did my own. After running your code on the 20 Newsgroups data set and training a simple classifier on learned document representations, I got a quite low accuracy (5% in my experiment). In addition, I found when I set max_iter too large, I got NaN errors when running the code.

Any help will be highly appreciated！

Thanks, Hugo

Contributor guide

Research direction
Investigate the cause of low classification accuracy and NaN errors. Check the NVDM model's training loop, hyperparameters (e.g., max iter), data preprocessing for 20 Newsgroups, and the classifier's learning process. Look for potential issues in the variational inference or reconstruction loss that might lead to degenerate representations.
Tech stack
pythontensorflow
Domain
machine learning
Issue type
Bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Needs investigation
Prerequisites
PythonTensorFlowBasic NLP concepts
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
60