arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2304.03376cs.LGmath.DSq-bio.NCq-bio.QM
收藏

神经群体动力学与几何的可解释统计表示

Interpretable statistical representations of neural population dynamics and geometry

EPFL(洛桑联邦理工学院);University Hospital Würzburg(维尔茨堡大学医院);Imperial College London(伦敦帝国学院);Blue Brain Project, EPFL(蓝脑计划(洛桑联邦理工学院));Medical University of Vienna(维也纳医科大学)

Adam Gosztolai, Robert L. Peach, Alexis Arnaudon, Mauricio Barahona, Pierre Vandergheynst

AI总结:

提出MARBLE方法,通过无监督几何深度学习将流形上的神经动力学分解为局部流场并映射到潜在空间,发现了跨网络和动物一致的潜在表示,实现了最先进的解码准确性。

摘要:

神经元群体的动力学通常在低维流形上演化。因此,我们需要学习神经流形上动力学过程的方法,以推断可解释且一致的潜在表示。我们引入了一种表示学习方法MARBLE,它将流形上的动力学分解为局部流场,并使用无监督几何深度学习将其映射到共同的潜在空间中。在模拟非线性动力系统、循环神经网络以及来自灵长类和啮齿类动物的实验性单神经元记录中,我们发现了涌现的低维潜在表示,这些表示参数化了增益调制、决策和内部状态变化期间的高维神经动力学。这些表示在神经网络和动物之间保持一致,从而能够对认知计算进行稳健比较。广泛的基准测试表明,与当前的表示学习方法相比,MARBLE在动物内和跨动物解码准确性上达到了最先进的水平,且用户输入最少。我们的结果表明,流形结构为开发强大的解码算法和跨实验同化数据提供了强大的归纳偏置。

英文摘要:

The dynamics of neuron populations commonly evolve on low-dimensional manifolds. Thus, we need methods that learn the dynamical processes over neural manifolds to infer interpretable and consistent latent representations. We introduce a representation learning method, MARBLE, that decomposes on-manifold dynamics into local flow fields and maps them into a common latent space using unsupervised geometric deep learning. In simulated non-linear dynamical systems, recurrent neural networks, and experimental single-neuron recordings from primates and rodents, we discover emergent low-dimensional latent representations that parametrise high-dimensional neural dynamics during gain modulation, decision-making, and changes in the internal state. These representations are consistent across neural networks and animals, enabling the robust comparison of cognitive computations. Extensive benchmarking demonstrates state-of-the-art within- and across-animal decoding accuracy of MARBLE compared with current representation learning approaches, with minimal user input. Our results suggest that manifold structure provides a powerful inductive bias to develop powerful decoding algorithms and assimilate data across experiments.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑