Automatic decoding of input sinusoidal signal in a neuron model: High pass homomorphic filtering
Automatic decoding of input sinusoidal signal in a neuron model: High pass homomorphic filtering
复制标题
DOI：
10.1016/j.neucom.2018.03.007
复制
发表时间：
2018-05-31
期刊：
影响因子：
6
通讯作者：
Liberti，Micaela
中科院分区：
文献类型：
--
作者：
Orcioni，Simone;Paffi，Alessandra;Liberti，Micaela
一种解码从正弦输入到神经元模型输出尖峰序列的信息的处理技术是理解神经元系统编码原理的理想工具。作者已经提出了一种自动解码程序，它基于信噪比（SNR）计算的改进版本，需要了解自发（没有输入信号）和受刺激（有输入信号）的神经元活动。在这项工作中，开发了一种基于高通同态滤波的自动解码过程，其性能与改进的信噪比相当或更好。像大多数信噪比方法一样，不需要神经元自发活动的优点是简化了程序，减少了解码信息所需的数据量，并且可以应用于无法获得神经元自发活动的环境。
A processing technique for decoding the information transferred from a sinusoidal input to the output spike sequence of a neuron model is a desirable tool for understanding the encoding principles of neuronal systems. An automatic decoding procedure, already proposed by the authors, is based on an improved version of the Signal to Noise Ratio (SNR) calculation and requires a knowledge of both spontaneous (in absence of input signal) and stimulated (in presence of input signal) neuronal activities. In this work, an automatic decoding procedure based on high-pass homomorphic filtering is developed that provides performances comparable or better than that obtained with the improved SNR. The advantages of not requiring the neuronal spontaneous activities, as most SNR methods do, are a procedure simplification, a reduction of the amount of data needed to decode the information, and the possibility of application to contexts where the neuronal spontaneous activity is not available.