Why should we add early exits to neural networks?
Deep neural networks are generally designed as a stack of differentiable layers, in which a
prediction is obtained only after running the full stack. Recently, some contributions have …
prediction is obtained only after running the full stack. Recently, some contributions have …
Decentralized and distributed learning for AIoT: A comprehensive review, emerging challenges and opportunities
H Xu, KP Seng, LM Ang, J Smith - IEEE Access, 2024 - ieeexplore.ieee.org
The advent of the Artificial Intelligent Internet of Things (AIoT) has sparked a revolution in the
deployment of intelligent systems, driving the need for innovative data processing …
deployment of intelligent systems, driving the need for innovative data processing …
The staircase property: How hierarchical structure can guide deep learning
E Abbe, E Boix-Adsera, MS Brennan… - Advances in …, 2021 - proceedings.neurips.cc
This paper identifies a structural property of data distributions that enables deep neural
networks to learn hierarchically. We define the``staircase''property for functions over the …
networks to learn hierarchically. We define the``staircase''property for functions over the …
Revisiting locally supervised learning: an alternative to end-to-end training
Due to the need to store the intermediate activations for back-propagation, end-to-end (E2E)
training of deep networks usually suffers from high GPUs memory footprint. This paper aims …
training of deep networks usually suffers from high GPUs memory footprint. This paper aims …
Efficienttrain: Exploring generalized curriculum learning for training visual backbones
The superior performance of modern deep networks usually comes with a costly training
procedure. This paper presents a new curriculum learning approach for the efficient training …
procedure. This paper presents a new curriculum learning approach for the efficient training …
ProgFed: Effective, communication, and computation efficient federated learning by progressive training
Federated learning is a powerful distributed learning scheme that allows numerous edge
devices to collaboratively train a model without sharing their data. However, training is …
devices to collaboratively train a model without sharing their data. However, training is …
[PDF][PDF] Accelerating federated learning with split learning on locally generated losses
Federated learning (FL) operates based on model exchanges between the server and the
clients, and suffers from significant communication as well as client-side computation …
clients, and suffers from significant communication as well as client-side computation …
Backpropagation-free deep learning with recursive local representation alignment
Training deep neural networks on large-scale datasets requires significant hardware
resources whose costs (even on cloud platforms) put them out of reach of smaller …
resources whose costs (even on cloud platforms) put them out of reach of smaller …
Learning without feedback: Fixed random learning signals allow for feedforward training of deep neural networks
While the backpropagation of error algorithm enables deep neural network training, it
implies (i) bidirectional synaptic weight transport and (ii) update locking until the forward and …
implies (i) bidirectional synaptic weight transport and (ii) update locking until the forward and …
Online learned continual compression with adaptive quantization modules
We introduce and study the problem of Online Continual Compression, where one attempts
to simultaneously learn to compress and store a representative dataset from a non iid data …
to simultaneously learn to compress and store a representative dataset from a non iid data …