Skip to content

Latest commit

 

History

History
27 lines (23 loc) · 687 Bytes

fc_layer.md

File metadata and controls

27 lines (23 loc) · 687 Bytes

Full connected Layer

  • e.g. A very simple example:
    img
  1. Input: your data, e.g: using weather data (humidity,temperature,pressure) to predict if will rain tomorrow.
  2. Hidden layer.
  3. Output: predicted result.
  • The relationship between Dx and Mx.
    $$M_1 = W_{11}*D_1 + W_{12}*D_2 + W_{13}*D_3$$
    $$M_2 = W_{21}*D_1 + W_{22}*D_2 + W_{23}*D_3$$

  • Obviously, it's a matrix operation. Let $W$ denote the weight matrix, and $D$ denote the input data:

$$D=\begin{bmatrix} D_1 & D_2 & D_3 \end{bmatrix} , W=\begin{bmatrix} W_{11} & W_{21} \\\ W_{12} & W_{22} \\\ W_{13} & W_{23} \end{bmatrix}, M=\begin{bmatrix} M_1 & M_2 \end{bmatrix}$$