-
Embedded event based object detection with spiking neural network
Authors:
Jonathan Courtois,
Pierre-Emmanuel Novac,
Edgar Lemaire,
Alain Pegatoquet,
Benoit Miramond
Abstract:
The complexity of event-based object detection (OD) poses considerable challenges. Spiking Neural Networks (SNNs) show promising results and pave the way for efficient event-based OD. Despite this success, the path to efficient SNNs on embedded devices remains a challenge. This is due to the size of the networks required to accomplish the task and the ability of devices to take advantage of SNNs b…
▽ More
The complexity of event-based object detection (OD) poses considerable challenges. Spiking Neural Networks (SNNs) show promising results and pave the way for efficient event-based OD. Despite this success, the path to efficient SNNs on embedded devices remains a challenge. This is due to the size of the networks required to accomplish the task and the ability of devices to take advantage of SNNs benefits. Even when "edge" devices are considered, they typically use embedded GPUs that consume tens of watts. In response to these challenges, our research introduces an embedded neuromorphic testbench that utilizes the SPiking Low-power Event-based ArchiTecture (SPLEAT) accelerator. Using an extended version of the Qualia framework, we can train, evaluate, quantize, and deploy spiking neural networks on an FPGA implementation of SPLEAT. We used this testbench to load a state-of-the-art SNN solution, estimate the performance loss associated with deploying the network on dedicated hardware, and run real-world event-based OD on neuromorphic hardware specifically designed for low-power spiking neural networks. Remarkably, our embedded spiking solution, which includes a model with 1.08 million parameters, operates efficiently with 490 mJ per prediction.
△ Less
Submitted 25 June, 2024;
originally announced June 2024.
-
An Analytical Estimation of Spiking Neural Networks Energy Efficiency
Authors:
Edgar Lemaire,
Loic Cordone,
Andrea Castagnetti,
Pierre-Emmanuel Novac,
Jonathan Courtois,
Benoit Miramond
Abstract:
Spiking Neural Networks are a type of neural networks where neurons communicate using only spikes. They are often presented as a low-power alternative to classical neural networks, but few works have proven these claims to be true. In this work, we present a metric to estimate the energy consumption of SNNs independently of a specific hardware. We then apply this metric on SNNs processing three di…
▽ More
Spiking Neural Networks are a type of neural networks where neurons communicate using only spikes. They are often presented as a low-power alternative to classical neural networks, but few works have proven these claims to be true. In this work, we present a metric to estimate the energy consumption of SNNs independently of a specific hardware. We then apply this metric on SNNs processing three different data types (static, dynamic and event-based) representative of real-world applications. As a result, all of our SNNs are 6 to 8 times more efficient than their FNN counterparts.
△ Less
Submitted 24 October, 2022;
originally announced October 2022.
-
Design Space Exploration of Hardware Spiking Neurons for Embedded Artificial Intelligence
Authors:
Nassim Abderrahmane,
Edgar Lemaire,
Benoît Miramond
Abstract:
Machine learning is yielding unprecedented interest in research and industry, due to recent success in many applied contexts such as image classification and object recognition. However, the deployment of these systems requires huge computing capabilities, thus making them unsuitable for embedded systems. To deal with this limitation, many researchers are investigating brain-inspired computing, wh…
▽ More
Machine learning is yielding unprecedented interest in research and industry, due to recent success in many applied contexts such as image classification and object recognition. However, the deployment of these systems requires huge computing capabilities, thus making them unsuitable for embedded systems. To deal with this limitation, many researchers are investigating brain-inspired computing, which would be a perfect alternative to the conventional Von Neumann architecture based computers (CPU/GPU) that meet the requirements for computing performance, but not for energy-efficiency. Therefore, neuromorphic hardware circuits that are adaptable for both parallel and distributed computations need to be designed. In this paper, we focus on Spiking Neural Networks (SNNs) with a comprehensive study of information coding methods and hardware exploration. In this context, we propose a framework for neuromorphic hardware design space exploration, which allows to define a suitable architecture based on application-specific constraints and starting from a wide variety of possible architectural choices. For this framework, we have developed a behavioral level simulator for neuromorphic hardware architectural exploration named NAXT. Moreover, we propose modified versions of the standard Rate Coding technique to make trade-offs with the Time Coding paradigm, which is characterized by the low number of spikes propagating in the network. Thus, we are able to reduce the number of spikes while kee** the same neuron's model, which results in an SNN with fewer events to process. By doing so, we seek to reduce the amount of power consumed by the hardware. Furthermore, we present three neuromorphic hardware architectures in order to quantitatively study the implementation of SNNs. These architectures are derived from a novel funnel-like Design Space Exploration framework for neuromorphic hardware.
△ Less
Submitted 1 October, 2019;
originally announced October 2019.
-
Pretty Modular Symmetric Encryption (PMSE), compact algorithm for ''embedded cryptography'' with quite low computational cost
Authors:
Etienne Lemaire
Abstract:
Nowadays, the dataflux shared between IOT systems must be secured from 8-bits to 64-bits processors systems. Several symmetric cryptographic algorithm already exist such as AES (Advanced Encryption Standard), RC4, Blowfish, etc. In this work, we propose an 8-bits encryption algorithm, combining several ideas of standard symmetric encryption algorithms. The aim is to provide a efficient, modular an…
▽ More
Nowadays, the dataflux shared between IOT systems must be secured from 8-bits to 64-bits processors systems. Several symmetric cryptographic algorithm already exist such as AES (Advanced Encryption Standard), RC4, Blowfish, etc. In this work, we propose an 8-bits encryption algorithm, combining several ideas of standard symmetric encryption algorithms. The aim is to provide a efficient, modular and compact algorithm able to tend to truly random one-time-pad encryption quality even for large data flux. The algorithm combines the implementation of a divergent polynomial with variable coefficients for pseudo-random keys generation, variable bit swap** and bitwise operations on data, and the use of one or two passwords. The encryption has been evaluated statistically, tested for image encryption and compared with One-Time-Pad encryption on the same data. Three implementations have been tested respectively in C, Javascript and GNU Octave. Encryption time performances are compared with AES on a 8-bits architecture: the Arduino Uno (ATmega328) microcontroller. A new concept of self-decryptionable web encrypted object, called Blocksnet is also presented and tested as a potential application of the compact and embedded PMSE algorithm.
△ Less
Submitted 17 May, 2019;
originally announced May 2019.
