Publications related to Half-precision floating-point format

Towards General-Purpose Decentralized Computing with Permissionless Extensibility

Smart contracts have emerged as the most promising foundations for applications of the blockchain technology. Even though smart contracts are expected to serve as the backbone of the next-generation web, they have several limitations that hinder their wide ...

EPFL2024

Functional-Basis Analysis of Non-Stationary Signals in Modern Power Grids: Theory and Implementation in Embedded Systems

Alexandra Cameron Karpilow

Situational awareness strategies are essential for the reliable and secure operation of the electric power grid which represents critical infrastructure in modern society. With the rise of converter-interfaced renewable generation and the consequent shift ...

EPFL2024

Low-Power Artificial Neural Network Perceptron Based on Monolayer MoS2

Aleksandra Radenovic, Andras Kis, Mukesh Kumar Tripathi, Guilherme Migliato Marega, Zhenyu Wang

Machine learning and signal processing on the edge are poised to influence our everyday lives with devices that will learn and infer from data generated by smart sensors and other devices for the Internet of Things. The next leap toward ubiquitous electron ...

2022

ColTraIn: Co-located DNN training and inference

Mario Paulo Drumond Lages De Oliveira

Deep neural network inference accelerators are deployed at scale to accommodate online services, but face low average load because of service demand variability, leading to poor resource utilization. Unfortunately, reclaiming inference idle cycles is diffi ...

EPFL2020

High Performance Computing for gravitational lens modeling: Single vs double precision on GPUs and CPUs

Jean-Paul Richard Kneib, Markus Rexroth, Christoph Ernst René Schäfer

Strong gravitational lensing is a powerful probe of cosmology and the dark matter distribution. Efficient lensing software is already a necessity to fully use its potential and the performance demands will only increase with the upcoming generation of tele ...

ELSEVIER2020

Mixed-Precision Deep Learning Based on Computational Memory

Irem Boybat Kara, Evangelos Eleftheriou, Abu Sebastian

Deep neural networks (DNNs) have revolutionized the field of artificial intelligence and have achieved unprecedented success in cognitive tasks such as image and speech recognition. Training of large DNNs, however, is computationally intensive and this has ...

FRONTIERS MEDIA SA2020

ESSOP: Efficient and Scalable Stochastic Outer Product Architecture for Deep Learning

Irem Boybat Kara, Evangelos Eleftheriou, Abu Sebastian

Deep neural networks (DNNs) have surpassed human-level accuracy in a variety of cognitive tasks but at the cost of significant memory/time requirements in DNN training. This limits their deployment in energy and memory limited applications that require rea ...

IEEE2020

Lactate measurement by neurochemical profiling in the dorsolateral prefrontal cortex at 7T: accuracy, precision, and relaxation times

Pierre Magistretti, Lijing Xin, Masoumeh Dehghani Moghadam, Kim Do

Purpose This assesses the potential of measuring lactate in the human brain using three non-editing MRS methods at 7T and compares the accuracy and precision of the methods. Methods H-1 MRS data were measured in the right dorsolateral prefrontal cortex usi ...

Wiley2020

Review and Benchmarking of Precision-Scalable Multiply-Accumulate Unit Architectures for Embedded Neural-Network Processing

Christian Enz, Vincent Frédéric Camus

The current trend for deep learning has come with an enormous computational need for billions of Multiply-Accumulate (MAC) operations per inference. Fortunately, reduced precision has demonstrated large benefits with low impact on accuracy, paving the way ...

2019

Survey of Precision-Scalable Multiply-Accumulate Units for Neural-Network Processing

Christian Enz, Vincent Frédéric Camus

The current trend for deep learning has come with an enormous computational need for billions of Multiply-Accumulate (MAC) operations per inference. Fortunately, reduced precision has demonstrated large benefits with low impact on accuracy, paving the way ...

IEEE2019

Multi-ReRAM synapses for artificial neural network training

Elmira Shahrabi, Igor Krawczuk, Yusuf Leblebici, Irem Boybat Kara, Carlo Ricciardi, Cecilia Giovinazzo, Evangelos Eleftheriou, Abu Sebastian

Metal-oxide-based resistive memory devices (ReRAM) are being actively researched as synaptic elements of neuromorphic co-processors for training deep neural networks (DNNs). However, device-level non-idealities are posing significant challenges. In this wo ...

IEEE2019

Non-perturbative probability distribution function for cosmological counts in cells

Sergey Sibiryakov, Mikhail Ivanov

We present a non-perturbative calculation of the 1-point probability distribution function (PDF) for the spherically-averaged matter density field. The PDF is represented as a path integral and is evaluated using the saddle-point method. It factorizes into ...

IOP PUBLISHING LTD2019

Trading off Energy versus Accuracy in Modern Computing Systems : From Digital Circuit Design to Programming Techniques

Jérémy Lucien Maurice Schlachter

The slowdown of Moore's law, which has been the driving force of the electronics industry over the last 5 decades, is causing serious problem to Integrated Circuits (ICs) improvements. Technology scaling is becoming more and more complex and fabrication co ...

EPFL2018

Training DNNs with Hybrid Block Floating Point

Babak Falsafi, Martin Jaggi, Tao Lin, Mario Paulo Drumond Lages De Oliveira

The wide adoption of DNNs has given birth to unrelenting computing requirements, forcing datacenter operators to adopt domain-specific accelerators to train them. These accelerators typically employ densely packed full-precision floating-point arithmetic t ...

NEURAL INFORMATION PROCESSING SYSTEMS (NIPS)2018

Mixed-precision architecture based on computational memory for training deep neural networks

Irem Boybat Kara, Evangelos Eleftheriou, Abu Sebastian

Deep neural networks (DNN) have revolutionized the field of machine learning by providing unprecedented human-like performance in solving many real-world problems such as image or speech recognition. Training of large DNNs, however, is a computationally in ...

IEEE2018

Hierarchical Reversible Logic Synthesis Using LUTs

Giovanni De Micheli, Mathias Soeken

Today's rapid advances in the physical implementation of quantum computers demand for scalable synthesis methods in order to map practical logic designs to quantum architectures. We present a synthesis algorithm for quantum computing based on k-LUT network ...

IEEE2017

Towards a Compiler for Reals

Viktor Kuncak, Eva Darulova

Numerical software, common in scientific computing or embedded systems, inevitably uses a finite-precision approximation of the real arithmetic in which most algorithms are designed. In many applications, the roundoff errors introduced by finite-precision ...

2017

An Inertial Sensor-Based Method for Estimating the Athlete's Relative Joint Center Positions and Center of Mass Kinematics in Alpine Ski Racing

Kamiar Aminian, Benedikt Fasel

For the purpose of gaining a deeper understanding of the relationship between external training load and health in competitive alpine skiing, an accurate and precise estimation of the athlete's kinematics is an essential methodological prerequisite. This s ...

Frontiers Media Sa2017

A Low-power Carry Cut-Back Approximate Adder with Fixed-point Implementation and Floating-point Precision

Christian Enz, Vincent Frédéric Camus, Jérémy Lucien Maurice Schlachter

This paper introduces an approximate adder architecture based on a digital quasi-feedback technique called Carry Cut-Back in which high-significance stages can cut the carry propagation chain at lower-significance positions. This lightweight approach preve ...

Ieee2016

Approximate 32-Bit Floating-Point Unit Design with 53% Power-Area Product Reduction

Christian Enz, Vincent Frédéric Camus, Jérémy Lucien Maurice Schlachter

The floating-point unit is one of the most common building block in any computing system and is used for a huge number of applications. By combining two state-of-the-art techniques of imprecise hardware, namely Gate-Level Pruning and Inexact Speculative Ad ...

Ieee2016