GenerativeNeurosci_ML-TrDic: Transformer + Dice Loss Framework for Bidirectional In Vitro–In Vivo Neural Spike-Train Generation

This release archives the code and supporting repository structure for reproducing the analyses associated with:

Shimono, M. (2026). In Vitro to In Vivo: Bidirectional and High-Precision Generation of In Vitro and In Vivo Neuronal Spike Data. Algorithms, 19(4), 305. https://doi.org/10.3390/a19040305

This repository implements a Transformer + Dice loss framework for bidirectional neural-domain transfer between unpaired in vitro and in vivo multineuronal spike trains.

This release is intended to support reproducibility, reuse, benchmark comparison, and citation of the software implementation associated with the paper.

If you use this code, benchmark, dataset structure, evaluation procedure, or any modified version of this repository, please cite the peer-reviewed article above.

Full Changelog: https://github.com/ShimonoMLab/GenerativeNeurosci_ML-TrDic/commits/v1.0.0