Publications related to Equinox: Training (for Free) on a Custom Inference Accelerator

MOD2IR: High-Performance Code Generation for a Biophysically Detailed Neuronal Simulation DSL

Felix Schürmann, Pramod Shivaji Kumbhar, Omar Awile, Snehal Pramod Kumbhar, Ioannis Magkanaris

Advances in computational capabilities and large volumes of experimental data have established computer simulations of brain tissue models as an important pillar in modern neuroscience. Alongside, a variety of domain specific languages (DSLs) have been dev ...

ACM2023

Overflow-free compute memories for edge AI acceleration

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

Compute memories are memory arrays augmented with dedicated logic to support arithmetic. They support the efficient execution of data-centric computing patterns, such as those characterizing Artificial Intelligence (AI) algorithms. These architectures can ...

2023

Design Space Exploration for Partitioning Dataflow Program on CPU-GPU Heterogeneous System

Marco Mattavelli, Simone Casale Brunet, Aurélien François Gilbert Bloch

Dataflow programming is a methodology that enables the development of high-level, parametric programs that are independent of the underlying platform. This approach is particularly useful for heterogeneous platforms, as it eliminates the need to rewrite ap ...

SPRINGER2023

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

Automatic and high-precision acoustic emission-based structural health monitoring of concrete structures

Brice Tanguy Alphonse Lecampion, Seyyedmaalek Momeni

Acoustic emission (AE) monitoring is a useful technique to monitor the health of a structure continuously, helping to prevent potential failure. AE are elastic waves produced and emitted during fracture processes inside a material and are recorded by senso ...

DESTech publications inc.2023

Beyond Spectral Gap: The Role of the Topology in Decentralized Learning

Martin Jaggi, Thijs Vogels, Hadrien Hendrikx

In data-parallel optimization of machine learning models, workers collaborate to improve their estimates of the model: more accurate gradients allow them to use larger learning rates and optimize faster. In the decentralized setting, in which workers commu ...

Microtome Publishing2023

Efficient Concurrent Analytical Query Processing using Data and Workload-conscious Sharing

Panagiotis Sioulas

Analytical workloads are evolving as the number of users surges and applications that submit queries in batches become popular. However, traditional analytical databases that optimize-then-execute each query individually struggle to provide timely response ...

EPFL2023

Power Transformer Fault Diagnosis Using Neural Network Optimization Techniques

Dimitrios Kaminaris

Artificial Intelligence (AI) techniques are considered the most advanced approaches for diagnosing faults in power transformers. Dissolved Gas Analysis (DGA) is the conventional approach widely adopted for diagnosing incipient faults in power transformers. ...

MDPI2023

Stable Nonconvex-Nonconcave Training via Linear Interpolation

Volkan Cevher, Thomas Michaelsen Pethick, Wanyun Xie

This paper presents a theoretical analysis of linear interpolation as a principled method for stabilizing (large-scale) neural network training. We argue that instabilities in the optimization process are often caused by the nonmonotonicity of the loss lan ...

2023

A 16-bit Floating-Point Near-SRAM Architecture for Low-power Sparse Matrix-Vector Multiplication

David Atienza Alonso, Giovanni Ansaloni, Grégoire Axel Eggermann, Marco Antonio Rios

State-of-the-art Artificial Intelligence (AI) algorithms, such as graph neural networks and recommendation systems, require floating-point computation of very large matrix multiplications over sparse data. Their execution in resource-constrained scenarios, ...

New York2023