SSSylvain Saïghi

Sylvain Saïghi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Sylvain Saïghi has a total of 24 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Line-based event preprocessing: towards low-energy neuromorphic computer vision

    by Amélie Gruel, Pierre Lewden, Adrien F Vincent, Sylvain Saïghi

    Neuromorphic vision has made significant progress in recent years, thanks to the natural match between spiking neural networks and event data in terms of biological inspiration, energy savings, latency and memory use for dynamic visual data processing. However, optimising its energy requirements still remains a challenge within the community, especially for embedded applications. One solution may reside in preprocessing events to optimise data quantities thus lowering the energy cost of neuromorphic hardware, proportional to the number of synaptic operations. To this end, we extend an end-to-end neuromorphic line detection mechanism to introduce line-based event data preprocessing. Our results demonstrate on three benchmark event-based datasets that preprocessing leads to an advantageous trade-off between energy consumption and classification performance. Depending on the line-based preprocessing strategy and the complexity of the classification task, we show that one can maintain or increase the classification accuracy while significantly reducing the theoretical energy consumption. Our approach systematically leads to a significant improvement of the neuromorphic classification efficiency, thus laying the groundwork towards a more frugal neuromorphic computer vision thanks to event preprocessing.


  • Object Detection with Spiking Neural Networks on Event Videos for Driving Scenarios

    by Shengyu Gao, Ying Tao, Amelie Gruel, Sylvain Saïghi, Chip-Hong Chang

    Event cameras excel in imaging objects under challenging lighting conditions with low latency. Their sparse and asynchronous event outputs closely align with the operational paradigm of Spiking Neural Networks (SNNs). An SNN mimics biological neurons of brain by transmitting information via binary spikes. The combination of biologically inspired event-based sensing with neuromorphic computing, while individually promising, faces significant challenges in their coherent integration to deliver better performance than existing solutions for real-time multi-scale, multi-rate object detection problems. In this paper, we propose a new Spike Tensor-based Multi-scale Object Detector (STMOD) for event video analytics in driving scenarios. STMOD features a cross-attention spatio-temporal spikes generator to convert asynchronous events into a high-dimension binary vector, a spiking backbone with three pyramid feature levels to fuse multi-scale spatio-temporal features and an asymmetric decoupled detection head to achieve high precision, energy efficient and responsive object detection without compromising the temporal resolution. Additionally, a new training method is proposed to overcome the mismatch in SNN’s state between training with independent samples and testing with video streaming to improve its performance on continuous video feed without impacting inference time resolution. Experimental results on the GEN1 and DSEC-Detection datasets show that STMOD achieves state-of-the-art mAP of 44.3% and 34.2%, respectively, with an inference time resolution of 5 ms. Finally, it also rivals existing event-based object detectors with significantly lower number of parameters and higher energy efficiency.


  • Open Access

    A ferroelectric–memristor memory for both training and inference

    by Michele Martemucci, François Rummens, Yannick Malot, Tifenn Hirtzlin, Olivier Guille, Simon Martin, Catherine Carabasse, Adrien F. Vincent, Sylvain Saïghi, Laurent Grenouillet, et al.

    Developing artificial intelligence systems that are capable of learning at the edge of a network requires both energy-efficient inference and learning. However, current memory technology cannot provide the necessary combination of high endurance, low programming energy and non-destructive read processes. Here we report a unified memory stack that functions as a memristor as well as a ferroelectric capacitor. Memristors are ideal for inference but have limited endurance and high programming energy; ferroelectric capacitors are ideal for learning, but their destructive read process makes them unsuitable for inference. Our memory stack uses a silicon-doped hafnium oxide and titanium scavenging layer that are integrated into the back end of line of a complementary metal–oxide–semiconductor process. With this approach, we fabricate an 18,432-device hybrid array (consisting of 16,384 ferroelectric capacitors and 2,048 memristors) with on-chip complementary metal–oxide–semiconductor periphery circuits. Each weight is associated with an analogue value stored as conductance levels in the memristors and a high-precision hidden value stored as a signed integer in the ferroelectric capacitors. Weight transfers between the different memory technologies occur without a formal digital-to-analogue converter. We use the array to validate an on-chip learning solution that, without batching, performs competitively with floating-point-precision software models across several benchmarks.


  • Open Access

    On-the-Fly Learning With Mixed-Mode Spiking Neural Network and Passive Memristive Array: Application to Neuromorphic Cameras

    by Pierre Lewden, Adrien F. Vincent, Jean Tomas, Sylvain Saïghi

    The massive deployment of the Internet of Things (IoT) combined with the need to reduce its impact on global energy consumption calls for the design of intelligent sensors. These sensors must be able to process information in situ to reduce the amount of data to be transferred, and to perform computing at low energy cost. Therefore, the design of intelligent sensors requires a compromise between performance and energy consumption. Spiking neural networks (SNNs) are good candidates for achieving the goal of an efficient intelligent sensor, as they use event-based computation. Interest for hardware implementation of SNNs has bloomed since the first experimental observation of memristors in 2008. In this paper, we study, by simulation means, the impact of the main technological parameters of Ferroelectric Tunnel Junctions (FTJs) synapses and of analog Leaky Integrate-and-Fire (LIF) neurons on the system learning capabilities. This allows us to determine which parameters are critical for the design of such systems and to suggest mitigation solutions as well as guidelines on how to build a SNN-based smart vision sensor in the context of unsupervised or reward-modulated learning. In particular, we show that splitting up the passive crossbar array of memristors could help dealing with the detrimental effect of the input voltage offset of the postsynaptic neurons.


  • Open Access

    A High-Level Methodology to Evaluate and Optimize Digital Architectures Targeting Spike Encoding

    by Clémence Gillet, Adrien F. Vincent, Bertrand Le Gal, Sylvain Saïghi

    Spiking Neural Networks (SNNs) are promising candidates for low-power and low-latency embedded artificial intelligence. However, those networks require event-based data produced by neuromorphic sensors which are not widely available, except for a few specialized devices like neuromorphic retinas. For other data types, a solution lies in the use of conventional sensors in conjunction with encoding layers. However, when performed in software, this solution can be detrimental to energy consumption or latency. Here we introduce a flexible design methodology for efficiently implementing, optimizing, and evaluating digital architectures of spike encoding integrating algorithms available in the literature. In order to quickly evaluate different hardware architectures and to tailor the solution to the application needs, our approach relies on High-Level Synthesis (HLS) tools and Python scripting. We illustrate the methodology by generating various digital architectures of two encoding algorithms taken from the literature and we evaluate their energy consumption and timing performances on Field Programmable Gate Arrays. This work could overcome the lack of neuromorphic sensors and accelerate the development of lower-power hardware SNNs.


  • Adjusting the neurons models in neuromimetic ICs using the voltage-clamp technique

    by Sylvain Saighi, Laure Buhry, Yannick Bornat, Gilles N'Kaoua, Jean Tomas, Sylvie Renaud

    This paper presents an original method to tune a neuromimetic IC based on neuron conductance-based models (Hodgkin-Huxley formalism). This method is well known in electrophysiology as the "voltage-clamp" technique. It consists in measuring the neural conductances while holding the membrane voltage at a clamped level. The voltage- and time- dependent variables of the conductance equations are then interpolated from the measurements. We apply this technique on the neuromimetic IC to extract the exact parameters of the neuron model. This model is computed in software, and results are compared with the hardware simulation. We conclude by mentioning the potential applications of this technique in hardware simulation systems based on neuromimetic ASICs.