File size: 3,100 Bytes
35f9b6e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 | ---
license: cc-by-4.0
tags:
- neuroscience
- computational-neuroscience
- neuroai
- generative-neuroscience
- spike-train
- spike-train-generation
- neural-data
- in-vitro
- in-vivo
- domain-transfer
- neural-domain-transfer
- transformer
- dice-loss
- sparse-time-series
- population-activity
- translational-neuroscience
papers:
- https://arxiv.org/abs/2503.20841
---
# In Vitro–In Vivo Spike-Train Generation Benchmark
This Hugging Face Dataset page provides a public entry point for the benchmark framework associated with the following paper:
**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**
Hugging Face Paper Page / arXiv preprint:
https://arxiv.org/abs/2503.20841
## Overview
This project introduces a benchmark framework for bidirectional neural-domain transfer between unpaired in vitro and in vivo multineuronal spike trains.
The central question is whether spontaneous population activity recorded in vitro can be transformed into in vivo-like neural activity, and whether in vivo activity can likewise be transformed into in vitro-like activity.
Because independent in vitro and in vivo recordings usually do not contain one-to-one matched neurons, the task is formulated as time-resolved neural-domain transfer between unpaired sparse binary spike-train sequences.
## Key Features
- Bidirectional in vitro-to-in-vivo and in vivo-to-in-vitro spike-train generation
- Sparse binary multineuronal time-series modeling
- 1-ms binned 128-unit spike-train representation
- Autoregressive Transformer model
- Dice loss for extreme spike-event sparsity
- ROC-AUC, Precision–Recall curves, and PR-AUC / average precision evaluation
- Benchmark concept for generative neuroscience and neural-domain transfer
## Code
The analysis code is available here:
https://github.com/ShimonoMLab/GenerativeNeurosci_ML-TrDic
## Data
The dataset associated with the paper is available via Mendeley Data:
https://doi.org/10.17632/kf65cvmtbz.1
## Recommended Citation
If you use the code, dataset structure, benchmark concept, preprocessing procedure, evaluation procedure, or any modified version of the repository, please cite:
**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**
Preprint:
**Shimono, M. (2025). _In vitro 2 In vivo: Bidirectional and High-Precision Generation of In Vitro and In Vivo Neuronal Spike Data_. arXiv:2503.20841. https://arxiv.org/abs/2503.20841**
## Suggested Citation Sentence
Shimono introduced a Transformer + Dice loss framework for bidirectional neural-domain transfer between unpaired in vitro and in vivo multineuronal spike trains.
## Keywords
generative neuroscience; computational neuroscience; NeuroAI; spike-train generation; sparse neural event prediction; Transformer; Dice loss; in vitro; in vivo; neural-domain transfer; translational neuroscience |