Publications related to A Flexible In-Memory Computing Architecture for Heterogeneously Quantized CNNs

Exploring High-Performance and Energy-Efficient Architectures for Edge AI-Enabled Applications

The desire and ability to place AI-enabled applications on the edge has grown significantly in recent years. However, the compute-, area-, and power-constrained nature of edge devices are stressed by the needs of the AI-enabled applications, due to a gener ...

EPFL2024

In-Sensor Passive Speech Classification with Phononic Metamaterials

Luis Guillermo Villanueva Torrijo, Daniel Moreno Garcia

Mitigating the energy requirements of artificial intelligence requires novel physical substrates for computation. Phononic metamaterials have vanishingly low power dissipation and hence are a prime candidate for green, always-on computers. However, their u ...

Wiley-V C H Verlag Gmbh2024

Solar-assisted clean steam generator, a cleaner production approach for sustainable healthcare facilities

Fernando Porté Agel, Ahmad Arabkoohsar, Mohammad Jamshidmofid

Healthcare sector buildings, particularly hospitals, significantly contribute to global energy consumption and greenhouse gas emissions. Hospitals require a substantial amount of steam, including the essential hygienic or " clean steam, " that is currently ...

Elsevier Sci Ltd2024

DBFS: Dynamic Bitwidth-Frequency Scaling for Efficient Software-defined SIMD

Giovanni Ansaloni, Alexandre Sébastien Julien Levisse, Flavio Ponzina, Pengbo Yu

Machine learning algorithms such as Convolutional Neural Networks (CNNs) are characterized by high robustness towards quantization, supporting small-bitwidth fixed-point arithmetic at inference time with little to no degradation in accuracy. In turn, small ...

IEEE2024

Teaching Not To

Charlotte Malterre-Barthes

Two design studios were successively taught in two schools of architecture: one at the Harvard Graduate School of Design (GSD) and the other at the Swiss Federal Institute of Technology in Lausanne (EPFL) in Lausanne, Switzerland, based on "A Global Morato ...

Taylor & Francis Inc2024

Electrical and Optical Manifestations of Flat Band Physics in Van der Waals Materials

Gabriele Pasquale

The scientific progress is significantly transforming contemporary society with the introduction and widespread application of technologies like artificial intelligence and quantum computing. Despite their profound impact, these technologies necessitate en ...

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

Co-Design Optimisation of Morphing Topology and Control of Winged Drones

Dario Floreano, Valentin Wüest, Fabio Bergonti

The design and control of winged aircraft and drones is an iterative process aimed at identifying a compromise of mission-specific costs and constraints. When agility is required, shape-shifting (morphing) drones represent an efficient solution. However, m ...

IEEE2024

Self-catalysed frontal polymerisation enables fast and low-energy processing of fibre reinforced polymer composites

Véronique Michaud, Baris Çaglar, Jacobus Gerardus Rudolph Staal

Frontal polymerisation has the potential to bring unprecedented reductions in energy demand and process time to produce fibre reinforced polymer composites. Production of epoxy-based fibre reinforced polymer parts with high fibre volume content, commonly e ...

2024

Climate risk assessment of buildings: An analysis of operating emissions of commercial offices in Australia

Arianna Brambilla

Building climate risk assessment involves benchmarking a building's energy use intensity against decarbonisation pathways to mitigate the impacts on climate change. Various climate risk assessment tools and frameworks are used for commercial buildings in d ...

Elsevier Science Sa2024

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

Robust NAS under adversarial training: benchmark, theory, and beyond

Volkan Cevher, Grigorios Chrysos, Fanghui Liu, Yongtao Wu

Recent developments in neural architecture search (NAS) emphasize the significance of considering robust architectures against malicious data. However, there is a notable absence of benchmark evaluations and theoretical guarantees for searching these robus ...

2024

Random matrix methods for high-dimensional machine learning models

Antoine Philippe Michel Bodin

In the rapidly evolving landscape of machine learning research, neural networks stand out with their ever-expanding number of parameters and reliance on increasingly large datasets. The financial cost and computational resources required for the training p ...

EPFL2024

An explorative method to support design decisions based on carbon constraints and daylight sufficiency needs

Nazanin Rezaei Oghazi

The global construction industry contributes to 37% of carbon emissions associated to both building operations and construction. To help achieve the net-zero targets set by 2050, it is mandated to achieve a 50% reduction in carbon emissions by 2030. As we ...

EPFL2024

Error assessment of an adaptive finite elements-neural networks method for an elliptic parametric PDE

Alexandre Caboussat, Marco Picasso, Maude Girardin

We present a finite elements-neural network approach for the numerical approximation of parametric partial differential equations. The algorithm generates training data from finite element simulations, and uses a data -driven (supervised) feedforward neura ...

Elsevier Science Sa2024

Hardware-Software co-design Methodologies for Edge AI Optimization

Flavio Ponzina

ML-based edge devices may face memory and computational errors that affect applications' reliability and performance. These errors can be the result of particular working conditions (e.g., radiation areas in physical experiments or avionics) or could be th ...

EPFL2023

A modelling framework for modern heritage buildings energy simulation

Franz Graf, Giulia Marino, Giuseppe Galbiati

The latest European polices highlight the urgent need to rehabilitate the existing building stock, responsible for 40 % of the EU's total energy consumption. In this process, a key role is played by thermal simulations, assessing the effective energy perfo ...

Elsevier2023

Building ventilation and outdoor air pollution: Implications for health, comfort and energy

Evangelos Belias

As humans spend most of their time indoors, indoor air quality (IAQ) significantly impacts their health. In parallel, building ventilation consumes significant energy, contributing to climate change. However, the relationships between the building ventilat ...

EPFL2023

A systemic study for enhanced waste heat recovery and renewable energy integration towards decarbonizing the aluminium industry

François Maréchal, Daniel Alexander Florez Orrego, Réginald Germanier

The aluminium remelting industry relies on natural gas to transform recycled aluminium into aluminium feedstock, entailing significant atmospheric emissions. Hydric resources are also affected as they are used as sinks of waste heat from the casting proces ...

2023

PePTM: An Efficient and Accurate Personalized P2P Learning Algorithm for Home Thermal Modeling

Rachid Guerraoui, Omid Ardakanian, Alexandre David Olivier Maurer

Nowadays, the integration of home automation systems with smart thermostats is a common trend, designed to enhance resident comfort and conserve energy. The introduction of smart thermostats that can run machine learning algorithms has opened the door for ...

MDPI2023