Aggelos K. Katsaggelos

TL;DR: It is shown that a high performing setup used in DNNs introduces a separation of time-scales in the training dynamics, allowing SGD to train layers from the lowest to the highest, and that for each layer, the distribution of solutions can be estimated using a class-based principal component analysis (PCA) of the layer's input.

TL;DR: This work addresses the problem of content-aware, foresighted resource reciprocation for media streaming over peer-to-peer (P2P) networks by introducing an artificial currency in order to maximize the video quality in the entire network.

TL;DR: A new quality metric to estimate the impact of packet loss on the perceptual quality of encoded video sequences transmitted over error-prone networks, and a No-Reference (NR) sparse regression model is presented to predict the proposed CDSSIM metric using pre-defined features associated with slice loss.

TL;DR: To improve the representative capabilities of existing spiking neurons, a novel neuron model is proposed called “location spiking neuron”, which enables us to extract features of event-based data in a novel way and a hybrid model which combines an SNN with TSRM neurons and an Snn with LSRM neuron to capture the complex spatio-temporal dependencies in the data is proposed.