# Re-balancing Variational Autoencoder Loss for Molecule Sequence Generation

arXiv: https://arxiv.org/abs/1910.00698 · HTML: https://mlsanity.com/paper/1910.00698/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 1 Oct 2019.

## Built on by (most cited first, 12 of 12)

- [Finetuning Pretrained Transformers into Variational Autoencoders](https://mlsanity.com/paper/2108.02446/) (2021)
- [Hybrid quantum-classical machine learning for generative chemistry and drug design](https://mlsanity.com/paper/2108.11644/) (2021)
- [Conditional β-VAE for De Novo Molecular Generation](https://mlsanity.com/paper/2205.01592/) (2022)
- [Latent Space Bayesian Optimization with Latent Data Augmentation for Enhanced Exploration](https://mlsanity.com/paper/2302.02399/) (2023)
- [String-based Molecule Generation via Multi-decoder VAE](https://mlsanity.com/paper/2208.10718/) (2022)
- [A Controllable Appearance Representation for Flexible Transfer and Editing](https://mlsanity.com/paper/2504.15028/) (2025)
- [Conditional Latent Space Molecular Scaffold Optimization for Accelerated Molecular Design](https://mlsanity.com/paper/2411.01423/) (2024)
- [AGMA: Adaptive Gaussian Mixture Anchors for Prior-Guided Multimodal Human Trajectory Forecasting](https://mlsanity.com/paper/2602.04204/) (2026)
- [High-Dimensional Bayesian Optimization with Constraints: Application to Powder Weighing](https://mlsanity.com/paper/2206.05988/) (2022)
- [Bottleneck-based Encoder-decoder ARchitecture (BEAR) for Learning Unbiased Consumer-to-Consumer Image Representations](https://mlsanity.com/paper/2409.06187/) (2024)
- [Detoxify: A framework for abusive text transformation using LLMs](https://mlsanity.com/paper/2507.10177/) (2025)
- [A Systematic Evaluation of Molecule Generation Models for De Novo Drug Design: From Benchmarks to Practical Insights](https://mlsanity.com/paper/2609.10099/) (2026)

## Abstract (arXiv)

Molecule generation is to design new molecules with specific chemical properties and further to optimize the desired chemical properties. Following previous work, we encode molecules into continuous vectors in the latent space and then decode the vectors into molecules under the variational autoencoder (VAE) framework. We investigate the posterior collapse problem of current RNN-based VAEs for molecule sequence generation. For the first time, we find that underestimated reconstruction loss leads to posterior collapse, and provide both theoretical and experimental evidence. We propose an effective and efficient solution to fix the problem and avoid posterior collapse. Without bells and whistles, our method achieves SOTA reconstruction accuracy and competitive validity on the ZINC 250K dataset. When generating 10,000 unique valid SMILES from random prior sampling, it costs JT-VAE1450s while our method only needs 9s. Our implementation is at https://github.com/chaoyan1037/Re-balanced-VAE.