Deep Neural Networks for Acoustic Modeling in Speech Recognition
Table of Contents

关于

深度学习在语音识别中的应用。参考文献

语音识别现状

$$
E(v, h) = \sum_{i \in vis} \frac{(v_i - a_i)^2}{2 \sigma_i^2} - \sum_{j \in hid} b_j h_j - \sum_{i,j} \frac{v_i}{\sigma_i} h_j w_{ij}
$$

两个条件分布为:

$$
p(h_j|v) = \text{logistic}(b_j + \sum_i \frac{v_i}{\sigma_i} w_{ij}) \\
p(v_i|h) = \mathcal{N}(a_i + \sigma_i \sum_j h_j w_{ij}; \sigma_i^2)
$$