Publications associées à Large-scale neural networks implemented with non-volatile memory as the synaptic weight element: impact of conductance response

Generalization of Scaled Deep ResNets in the Mean-Field Regime

Volkan Cevher, Grigorios Chrysos, Fanghui Liu

Despite the widespread empirical success of ResNet, the generalization properties of deep ResNet are rarely explored beyond the lazy training regime. In this work, we investigate scaled ResNet in the limit of infinitely deep and wide neural networks, of wh ...

2024

Extensions of Peer Prediction Incentive Mechanisms

Adam Julian Richardson

As large, data-driven artificial intelligence models become ubiquitous, guaranteeing high data quality is imperative for constructing models. Crowdsourcing, community sensing, and data filtering have long been the standard approaches to guaranteeing or imp ...

EPFL2024

Towards Scalable Electronics: Synthetic 2D Materials for Large-Area 2D Circuit Integration

Zhenyu Wang

In the past decades, a significant increase of the transistor density on a chip has led to exponential growth in computational power driven by Moore's law. To overcome the bottleneck of traditional von-Neumann architecture in computational efficiency, effo ...

EPFL2024

EdgeAI-Aware Design of In-Memory Computing Architectures

Marco Antonio Rios

Driven by the demand for real-time processing and the need to minimize latency in AI algorithms, edge computing has experienced remarkable progress. Decision-making AI applications stand out for their heavy reliance on data-centric operations, predominantl ...

EPFL2024

Instabilities in Elastic and Magneto-Elastic Beams and Shells: from Shell Buckling to the Design of a Programmable Braille Reader

Arefeh Abbasi

In this thesis, we conduct a comprehensive investigation into structural instabilities of both elastic and magneto-elastic beams and shells, resulting in a creative proposal to design a programmable braille reader. Methodologically, we combine numerical si ...

EPFL2024

Deep Learning Theory Through the Lens of Diagonal Linear Networks

Scott William Pesme

In this PhD manuscript, we explore optimisation phenomena which occur in complex neural networks through the lens of

2

-layer diagonal linear networks. This rudimentary architecture, which consists of a two layer feedforward linear network with a diagonal ...

EPFL2024

A story of two transitions: From adhesive to abrasive wear and from ductile to brittle regime

Jean-François Molinari, Sacha Zenon Wattel

Atomistic simulations performed with a family of model potential with tunable hardness have proven to be a great tool for advancing the understanding of wear processes at the asperity level. They have been instrumental in finding a critical length scale, w ...

2024

Deep learning approach for identification of H II regions during reionization in 21-cm observations - II. Foreground contamination

Michele Bianco, Jean-Paul Richard Kneib, Emma Elizabeth Tolley, Tianyue Chen

The upcoming Square Kilometre Array Observatory will produce images of neutral hydrogen distribution during the epoch of reionization by observing the corresponding 21-cm signal. However, the 21-cm signal will be subject to instrumental limitations such as ...

Oxford Univ Press2024

Chaosity: Understanding Contemporary NUMA-architectures

Anastasia Ailamaki, Viktor Sanca, Syed Mohammad Aunn Raza, Hamish Mcniece Hill Nicholson, Andreea Nica

Modern hardware is increasingly complex, requiring increasing effort to understand in order to carefully engineer systems for optimal performance and effective utilization. Moreover, established design principles and assumptions are not portable to modern ...

2023

HetCache: Synergising NVMe Storage and GPU acceleration for Memory-Efficient Analytics

Anastasia Ailamaki, Periklis Chrysogelos, Syed Mohammad Aunn Raza, Hamish Mcniece Hill Nicholson

Accessing input data is a critical operation in data analytics: i) slow data access significantly degrades performance, and ii) storing everything in the fastest medium, i.e., memory, incurs high operational and hardware costs. Further, while GPUs offer in ...

2023

2D Nanosystems: Applications of 2D Semiconductors for In-Memory Computing

Guilherme Migliato Marega

Machine learning and data processing algorithms have been thriving in finding ways of processing and classifying information by exploiting the hidden trends of large datasets. Although these emerging computational methods have become successful in today's ...

EPFL2023

Towards Robust Monitoring of the Laser Powder Bed Fusion Process based on Acoustic Emission combined with Machine Learning Solutions

Rita Drissi Daoudi

Laser Powder Bed Fusion (LPBF) is an Additive Manufacturing (AM) process consolidating parts layer by layer, from a metallic powder bed. It allows no limitation in terms of geometry and is therefore of particular interest to various industries. Metallic LP ...

EPFL2023

Polynomial-time universality and limitations of deep learning

Emmanuel Abbé

The goal of this paper is to characterize function distributions that general neural networks trained by descent algorithms (GD/SGD), can or cannot learn in polytime. The results are: (1) The paradigm of general neural networks trained by SGD is poly-time ...

WILEY2023

Bit-Line Computing for CNN Accelerators Co-Design in Edge AI Inference

David Atienza Alonso, Giovanni Ansaloni, Alexandre Sébastien Julien Levisse, Flavio Ponzina, Marco Antonio Rios

By supporting the access of multiple memory words at the same time, Bit-line Computing (BC) architectures allow the parallel execution of bit-wise operations in-memory. At the array periphery, arithmetic operations are then derived with little additional o ...

2023

Learning Robust and Adaptive Representations: from Interactions, for Interactions

Yuejiang Liu

Interactions are ubiquitous in our world, spanning from social interactions between human individuals to physical interactions between robots and objects to mechanistic interactions among different components of an intelligent system. Despite their prevale ...

EPFL2023

Building Chips Faster: Hardware-Compiler Co-Design for Accelerated RTL Simulation

Sahand Kashani

The demise of Moore's Law and Dennard scaling has resulted in diminishing performance gains for general-purpose processors, and so has prompted a surge in academic and commercial interest for hardware accelerators. Specialized hardware has already redefine ...

EPFL2023

CONVERGENCE AND NONCONVERGENCE OF SCALED SELF-INTERACTING RANDOM WALKS TO BROWNIAN MOTION PERTURBED AT EXTREMA

Thomas Mountford

We use generalized Ray-Knight theorems, introduced by B. Toth in 1996, together with techniques developed for excited random walks as main tools for establishing positive and negative results concerning convergence of some classes of diffusively scaled sel ...

Inst Mathematical Statistics-Ims2023

Regressive Tomography and Direct Regressive Reconstruction: A Research Note

Henri Weisen

Two related methods for inverting line-integrated measurements are presented in this research paper in the context of the recent deuterium-tritium experiments in the JET tokamak. Unlike traditional methods of tomography, these methods rely on making use of ...

2023

Higher Order Asymptotics: Applications to Satellite Conjunction and Boundary Problems

Soumaya Elkantassi

Higher-order asymptotics provide accurate approximations for use in parametric statistical modelling. In this thesis, we investigate using higher-order approximations in two-specific settings, with a particular emphasis on the tangent exponential model. ...

EPFL2023

Rapid Network Adaptation: Learning to Adapt Neural Networks Using Test-Time Feedback

Amir Roshan Zamir, Shuqing Teresa Yeo, Oguzhan Fatih Kar, Zahra Sodagar

We propose a method for adapting neural networks to distribution shifts at test-time. In contrast to training-time robustness mechanisms that attempt to anticipate and counter the shift, we create a closed-loop system and make use of test-time feedback sig ...

Ieee Computer Soc2023