Fig. 1
From: Image processing with Optical matrix vector multipliers implemented for encoding and decoding tasks
Schematic of the optical MVM operations used in the ONN-based autoencoder. a General structure of a DNN utilizing FC-NNs with MVMs. The key operation, MVM, involves multiplying a weight matrix WN×K by an input vector XK, performing element-wise multiplications and accumulating the results, producing an output vector YN. b Representation of an autoencoder built with FC-NNs, where the encoder compresses high-dimensional input data into a lower-dimensional latent vector, and the decoder reconstructs this data. c Type 1 MVM operation showing the fan-out, multiplication, and fan-in processes for encoding tasks where the output dimensions are smaller than input dimensions (K >> N). d Type 2 MVM operation showing the enhanced strategy for decoding tasks, where the output dimensions exceed input dimensions (K << N)
