Real-time execution of SNN models with synaptic plasticity for handwritten digit recognition on SIMD hardware.

Recent advancements in neuromorphic computing have led to the development of hardware architectures inspired by Spiking Neural Networks (SNNs) to emulate the efficiency and parallel processing capabilities of the human brain. This work focuses on testing the HEENS architecture, specifically designed for high parallel processing and biological realism in SNN emulation, implemented on a ZYNQ family FPGA. The study applies this architecture to the classification of digits using the well-known MNIST database.

Real-time hardware emulation of neural cultures: A comparative study of in vitro, in silico and in duris silico models.

Biological neural networks are well known for their capacity to process information with extremely low power consumption. Fields such as Artificial Intelligence, with high computational costs, are seeking for alternatives inspired in biological systems. An inspiring alternative is to implement hardware architectures that replicate the behavior of biological neurons but with the flexibility in programming capabilities of an electronic device, all combined with a relatively low operational cost.

Resonant MEMS Pressure Sensor in 180 nm CMOS Technology Obtained by BEOL Isotropic Etching.

This work presents the design and characterization of a resonant CMOS-MEMS pressure sensor manufactured in a standard 180 nm CMOS industry-compatible technology. The device consists of aluminum square plates attached together by means of tungsten vias integrated into the back end of line (BEOL) of the CMOS process. Three prototypes were designed and the structural characteristics were varied, particularly mass and thickness, which are directly related to the resonance frequency, quality factor, and pressure; while the same geometry at the frontal level, as well as the air gap, were maintained to allow structural comparative analysis of the structures.

A CMOS-MEMS BEOL 2-axis Lorentz-Force Magnetometer with Device-Level Offset Cancellation.

Lorentz-force Microelectromechanical Systems (MEMS) magnetometers have been proposed as a replacement for magnetometers currently used in consumer electronics market. Being MEMS devices, they can be manufactured in the same die as accelerometers and gyroscopes, greatly reducing current solutions volume and costs. However, they still present low sensitivities and large offsets that hinder their performance.

Experiments on MEMS Integration in 0.25 μm CMOS Process.

In this paper, we share our practical experience gained during the development of CMOS-MEMS (Complementary Metal-Oxide Semiconductor Micro Electro Mechanical Systems) devices in IHP SG25 technology. The experimental prototyping process is illustrated with examples of three CMOS-MEMS chips and starts from rough process exploration and characterization, followed by the definition of the useful MEMS design space to finally reach CMOS-MEMS devices with inertial mass up to 4.3 μg and resonance frequency down to 4.

Action potential propagation: ion current or intramembrane electric field?

The established action potential propagation mechanisms do not satisfactorily explain propagation on myelinated axons given the current knowledge of biological channels and membranes. The flow across ion channels presents two possible effects: the electric potential variations across the lipid bilayers (action potential) and the propagation of an electric field through the membrane inner part. The proposed mechanism is based on intra-membrane electric field propagation, this propagation can explain the action potential saltatory propagation and its constant delay independent of distance between Ranvier nodes in myelinated axons.

SNAVA-A real-time multi-FPGA multi-model spiking neural network simulation architecture.

Spiking Neural Networks (SNN) for Versatile Applications (SNAVA) simulation platform is a scalable and programmable parallel architecture that supports real-time, large-scale, multi-model SNN computation. This parallel architecture is implemented in modern Field-Programmable Gate Arrays (FPGAs) devices to provide high performance execution and flexibility to support large-scale SNN models. Flexibility is defined in terms of programmability, which allows easy synapse and neuron implementation.

Integration of GMR Sensors with Different Technologies.

Less than thirty years after the giant magnetoresistance (GMR) effect was described, GMR sensors are the preferred choice in many applications demanding the measurement of low magnetic fields in small volumes. This rapid deployment from theoretical basis to market and state-of-the-art applications can be explained by the combination of excellent inherent properties with the feasibility of fabrication, allowing the real integration with many other standard technologies. In this paper, we present a review focusing on how this capability of integration has allowed the improvement of the inherent capabilities and, therefore, the range of application of GMR sensors.