神经元的区分能力与置信水平(英文)

Neuronal Discriminability and Confidence Level

当神经元接受两种不同的刺激时,利用发放率分布考虑神经元的区分能力与置信水平,对其分布分别服从相同和不同方差的高斯分布情形,给出了度量两种可能分布的分开程度的量―区分能力,和度量这种区分可靠性的量―置信水平,刻画了区分能力与其置信水平之间的关系,并给出了最优阈值的选择以及执行最优区分任务的解码过程.

Neuronal discriininability and confidence level are considered by the distributions of firing rates when one neuron receives two different stimulus. Two Gaussian distributions of firing rates, both with the same variance and with different variance, are discussed. The discrimi＇nability to measure the separation between two possible distributions and the confidence level to measure the reliability of discrimination are given. The relationship between the discriminability and the confidence level is characterized. In addition, the optimal threshold is included. As a result, a decoding procedure to perform the discrimination task perfectly is proposed.