Deep neural networks (DNNs) have gained considerable attention in various real-world
applications due to the strong performance on representation learning. However, a DNN …

Despite its widespread adoption, Adam's advantage over Stochastic Gradient Descent
(SGD) lacks a comprehensive theoretical explanation. This paper investigates Adam's …

We study the training of deep neural networks by gradient descent where floating-point
arithmetic is used to compute the gradients. In this framework and under realistic …

Floating-Point (FP) units in processors are generally limited to supporting a subset of formats
defined by the IEEE 754 standard. As a result, high-efficiency languages and optimizing …

V magistrski nalogi realiziramo nevronsko mrežo večslojni perceptron in štiri različne učne
algoritme. Nevronska mreža in učni algoritmi so modularni kar pomeni, da lahko …

A system for increasing quality of results of computations of an artificial neural network
(ANN) by using complex rounding rules for parameters in the ANN is provided, the system …

V magistrskem delu smo razvili programsko ogrodje za realizacijo in pohitritev delovanja
polno povezanih nevronskih mrez na vezjih FPGA. Nevronske mreze so izvedene v visoko …