Neural Latents Benchmark '21

Neural Latents Benchmark '21是由佐治亚理工学院的研究团队开发的一个数据集，专注于评估神经群体活动的潜在变量模型。该数据集包含四个子集，分别来自认知、感觉和运动区域的神经尖峰活动记录。这些数据集的目的是推动开发能够适用于不同脑区和行为的广泛神经活动的模型。通过这些数据集，研究者可以评估和比较不同的潜在变量模型，从而推动神经科学领域的方法标准化和模型发展。Neural Latents Benchmark '21 is a dataset developed by a research team from the Georgia Institute of Technology, focused on evaluating latent variable models for neural population activity. This dataset includes four subsets sourced from neural spike recordings in cognitive, sensory, and motor brain regions. The purpose of these datasets is to promote the development of models that can generalize across diverse neural activity patterns across different brain regions and behavioral scenarios. Using these datasets, researchers can evaluate and compare various latent variable models, thereby advancing methodological standardization and model development in the field of neuroscience.

搜集汇总

背景与挑战

背景概述

Neural Latents Benchmark '21（NLB 2021）是一个用于评估神经尖峰活动潜在变量模型的基准套件，包含7个数据集，覆盖4个不同任务和脑区，旨在促进神经数据分析的建模工作。该基准最初通过EvalAI平台进行模型提交和评估，但自2026年1月起已停止接受新提交，并公开了测试数据以供本地评估。