3 ms·

Neural nets are structured as matrix multiplication, yet, they are universal approximators.

• noosphr 9mo ago
  You're missing the non-linear activations.