Publications associées à Netscope: Practical Network Loss Tomography

Communication-efficient distributed training of machine learning models

In this thesis, we explore techniques for addressing the communication bottleneck in data-parallel distributed training of deep learning models. We investigate algorithms that either reduce the size of the messages that are exchanged between workers, or th ...

EPFL2023

Time vs. Truth: Age-Distortion Tradeoffs and Strategies for Distributed Inference

Yunus Inan

In 1948, Claude Shannon laid the foundations of information theory, which grew out of a study to find the ultimate limits of source compression, and of reliable communication. Since then, information theory has proved itself not only as a quest to find the ...

EPFL2023

Evaluating the effect of sparse convolutions on point cloud compression

Touradj Ebrahimi, Davi Nachtigall Lazzarotto

The use of point clouds as an imaging modality has been rapidly growing, motivating research on compression methods to enable efficient transmission and storage for many applications. While compression standards relying on conven- tional techniques such as ...

2023

Dual-frequency spectral radar retrieval of snowfall microphysics: a physics-driven deep-learning approach

Alexis Berne, Gionata Ghiggi, Anne-Claire Marie Billault--Roux

The use of meteorological radars to study snowfall microphysical properties and processes is well established, in particular via a few distinct techniques: the use of radar polarimetry, of multi-frequency radar measurements, and of the radar Doppler spectr ...

COPERNICUS GESELLSCHAFT MBH2023

Integrated electronics for time-of-flight positron emission tomography photodetectors

Andrada Alexandra Muntean

Positron emission tomography is a nuclear imaging technique well known for its use in oncology for cancer diagnosis and staging. A PET scanner is a complex machine which comprises photodetectors placed in a ring configuration that detect gamma photons gen ...

EPFL2023

Influence of Spatial Rendering on the Performance of Point Cloud Objective Quality Metrics

Touradj Ebrahimi, Michela Testolina

The use of point clouds to digitally represent three-dimensional objects with both geometry and color attributes is rapidly increasing in several applications. Since storage and transmission of uncompressed point cloud data are often impractical, several l ...

IEEE2022

Stochastic distributed learning with gradient quantization and double-variance reduction

Sebastian Urban Stich, Konstantin Mishchenko

We consider distributed optimization over several devices, each sending incremental model updates to a central server. This setting is considered, for instance, in federated learning. Various schemes have been designed to compress the model updates in orde ...

TAYLOR & FRANCIS LTD2022

Latent Space Slicing for Enhanced Entropy Modeling in Learning-Based Point Cloud Geometry Compression

Touradj Ebrahimi, Davi Nachtigall Lazzarotto

The growing adoption of point clouds as an imaging modality has stimulated the search for efficient solutions for compression. Learning-based algorithms have been reporting increasingly better performance and are drawing the attention from the research com ...

IEEE2022

Structured and tiled-based pruning of Deep Learning models targeting FPGA implementations

Alexandre Schmid, Lizeth Gonzalez Carabarin

Model compression techniques have lead to a reduction of size and number of computations of Deep Learning models. However, techniques such as pruning mostly lack of a real co-optimization with hardware platforms. For instance, implementing unstructured pru ...

IEEE2022

Masked Training of Neural Networks with Partial Gradients

Martin Jaggi, Sebastian Urban Stich, Amirkeivan Mohtashami

State-of-the-art training algorithms for deep learning models are based on stochastic gradient descent (SGD). Recently, many variations have been explored: perturbing parameters for better accuracy (such as in Extra-gradient), limiting SGD updates to a sub ...

JMLR-JOURNAL MACHINE LEARNING RESEARCH2022

Evaluation and optimization of novel extraction algorithms for the automatic detection of atrial activations recorded within the pulmonary veins during atrial fibrillation

Jean-Marc Vesin, Yann Prudat, Etienne Pruvot, Sasan Yazdani, Adrian Luca

Background and objective The automated detection of atrial activations (AAs) recorded from intracardiac electrograms (IEGMs) during atrial fibrillation (AF) is challenging considering their various amplitudes, morphologies and cycle length. Activation time ...

BMC2022

Smart pooling: AI-powered COVID-19 informative group testing

Jorge Andres Madrid Wolff

Massive molecular testing for COVID-19 has been pointed out as fundamental to moderate the spread of the pandemic. Pooling methods can enhance testing efficiency, but they are viable only at low incidences of the disease. We propose Smart Pooling, a machin ...

Nature Portfolio2022

OSLO: On-the-Sphere Learning for Omnidirectional Images and Its Application to 360-Degree Image Compression

Pascal Frossard, Thomas Maugey, Roberto Gerson De Albuquerque Azevedo

State-of-the-art 2D image compression schemes rely on the power of convolutional neural networks (CNNs). Although CNNs offer promising perspectives for 2D image compression, extending such models to omnidirectional images is not straightforward. First, omn ...

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC2022

Towards Verifiable, Generalizable and Efficient Robust Deep Neural Networks

Chen Liu

In the last decade, deep neural networks have achieved tremendous success in many fields of machine learning. However, they are shown vulnerable against adversarial attacks: well-designed, yet imperceptible, perturbations can make the state-of-the-art deep ...

EPFL2022

Learning, compression, and leakage: Minimising classification error via meta-universal compression principles

Michael Christoph Gastpar, Fernando Rosas de Andraca

Learning and compression are driven by the common aim of identifying and exploiting statistical regularities in data, which opens the door for fertile collaboration between these areas. A promising group of compression techniques for learning scenarios is ...

IEEE2021

Large-Scale Crowdsourcing Subjective Quality Evaluation of Learning-Based Image Coding

Touradj Ebrahimi, Evgeniy Upenik, Michela Testolina

Learning-based image codecs produce different compression artifacts, when compared to the blocking and blurring degradation introduced by conventional image codecs, such as JPEG, JPEG~2000 and HEIC. In this paper, a crowdsourcing based subjective quality e ...

IEEE2021

Hardware-oriented pruning and quantization of Deep Learning models to detect life-threatening arrhythmias

Alexandre Schmid, Lizeth Gonzalez Carabarin

Wearable solutions based on Deep Learning (DL) for real-time ECG monitoring are a promising alternative to detect life-threatening arrhythmias. However, DL models suffer of a large memory footprint, which hampers their adoption in portable technologies. Th ...

IEEE2021

Perceptual Quality of Point Clouds with application to Compression

Evangelos Alexiou

Modern information technologies and human-centric communication systems employ advanced content representations for richer portrayals of the real world. The newly adopted imaging modalities offer additional information cues and permit the depiction of real ...

EPFL2021

Direction is what you need: Improving Word Embedding Compression in Large Language Models

Karl Aberer, Rémi Philippe Lebret, Mohammadreza Banaei

The adoption of Transformer-based models in natural language processing (NLP) has led to great success using a massive number of parameters. However, due to deployment constraints in edge devices, there has been a rising interest in the compression of thes ...

ASSOC COMPUTATIONAL LINGUISTICS-ACL2021

Partition and Code: learning how to compress graphs

Andreas Loukas, Georgios Bouritsas, Nikolaos Karalias, Michael Bronstein

Can we use machine learning to compress graph data? The absence of ordering in graphs poses a significant challenge to conventional compression algorithms, limiting their attainable gains as well as their ability to discover relevant patterns. On the other ...

2021