A5042805196- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Semiconductor materials and devices
- Advancements in Semiconductor Devices and Circuit Design
- Neural Networks and Reservoir Computing
- Neuroscience and Neural Engineering
- CCD and CMOS Imaging Sensors
- Ferroelectric and Piezoelectric Materials
- MXene and MAX Phase Materials
- Quantum Computing Algorithms and Architecture
- Neural Networks and Applications
- Advanced Data Storage Technologies
- Integrated Circuits and Semiconductor Failure Analysis
- Nonlinear Dynamics and Pattern Formation
- Transition Metal Oxide Nanomaterials
- Parallel Computing and Optimization Techniques
- Neural dynamics and brain function
- 3D IC and TSV technologies
- ZnO doping and properties
- Robotics and Sensor-Based Localization
- Speech and dialogue systems
- Computer Graphics and Visualization Techniques
- Near-Field Optical Microscopy
- Modular Robots and Swarm Intelligence
- Molecular Communication and Nanonetworks
META Health
2024-2025
Universidad de Los Andes, Chile
2024
University of Notre Dame
2018-2023
Pontificia Universidad Católica de Chile
2017-2020
STMicroelectronics (France)
2012
STMicroelectronics (Switzerland)
2011
The two possible pathways toward artificial intelligence (AI)-(i) neuroscience-oriented neuromorphic computing [like spiking neural network (SNN)] and (ii) computer science driven machine learning (like deep learning) differ widely in their fundamental formalism coding schemes (Pei et al., 2019). Deviating from traditional approach of relying on neuronal models with static nonlinearities, SNNs attempt to capture brain-like features like computation using spikes. This holds the promise...
We report the first experimental demonstration of ferroelectric field-effect transistor (FEFET) based spiking neurons. A unique feature (FE) neuron demonstrated herein is availability both excitatory and inhibitory input connections in compact 1T-1FEFET structure, which also reported for time any implementations. Such dual functionality a key requirement bio-mimetic neural networks represents breakthrough implementation third generation (SNNs)-also unsupervised learning clustering on real...
Abstract The striking similarity between biological locomotion gaits and the evolution of phase patterns in coupled oscillatory network can be traced to role central pattern generator located spinal cord. Bio-inspired robotics aim at harnessing this control approach for generation rhythmic synchronized limb movement. Here, we utilize phenomenon synchronization emergent spatiotemporal from interaction among oscillators generate a range gait patterns. We experimentally demonstrate using...
We experimentally demonstrate W-doped amorphous In <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> O xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> double-gate field-effect transistors (DG IWO FET) with 5nm channel thickness and 50nm length exhibiting (a) excellent subthreshold slope (SS) of 73mV/dec, (b) record ID,SAT 550μA/μm at V xmlns:xlink="http://www.w3.org/1999/xlink">GS</sub> -V...
Pseudo-crossbar arrays using ferroelectric field effect transistor (FEFET) mitigates weight movement and allows <i>in situ</i> vector–matrix multiplication (VMM), which can significantly accelerate online training of deep neural networks (DNNs). However, the accuracy DNNs conventional FEFETs is low because non-idealities, such as nonlinearity, asymmetry, limited bit precision, dynamic range updates. The endurance these devices degrades further. Here, we show a novel approach for...
We demonstrate that a ferroelectric field-effect transistor (FeFET)-based spiking neuron is capable of mimicking various and bursting patterns characteristic cortical neurons. propose compact model to describe the dynamical behavior such FeFET-based This captures current-voltage dynamics FeFET critical voltages its hysteretic region. It aimed at system-level modeling simulation biomimetic networks neurons are ideal for neuromorphic computing.
We present three possible extensions of the bit-flip channel to qutrit systems based on diverse interpretations channel. Also, we extended them higher-dimensional qudit systems, formulating different versions dit flip channels. Finally, studied their impact Negativity, as an entanglement measure, qubit-qutrit and 2-qutrit Werner states. In doing so, showed inequivalence these they affect in very distinct ways states entanglement.
Finding the ground state of an Ising model maps to certain classes combinatorial optimization problems. Currently, several physical systems, called machines, are being sought provide optimal solution this otherwise NP-hard problem. In work, we experimentally demonstrate: (a) artificial spin states using second harmonic injection locking (SHIL) in insulator-to-metal phase transition nano-oscillators (IMT-NOs), (b) anti-ferromagnetic (and ferromagnetic) coupling capacitive resistive) and (c)...
Off-chip DRAM memory accesses limit the energy efficiency and training time of state-of-the-art deep neural networks (DNN). Compute-in-memory (CIM) accelerators leveraging pseudo-crossbar arrays on-chip weight storage have emerged as alternatives to GPUs for fast efficient training. However, this comes at cost reduced accuracy due cell non-idealities such as: low bit precision, nonlinearity, asymmetry, G <inf xmlns:mml="http://www.w3.org/1998/Math/MathML"...
Deeper understanding of memristive behavior is the only safe way towards maximum exploitation favorable properties and analog nature this new device technology in innovative applications. This can be achieved through experimental hands-on experience with real devices. However, lab experiments memristors are a challenging step, especially for uninitiated. In direction, paper presents some important considerations to carry out reliable measurements using an setup composed <italic...
Biologically plausible mechanism like homeostasis compliments Hebbian learning to allow unsupervised in spiking neural networks [1]. In this work, we propose a novel ferroelectric-based quasi-LIF neuron that induces intrinsic homeostasis. We experimentally characterize and perform phase-field simulations delineate the non-trivial transient polarization relaxation associated with multi-domain interaction poly-crystalline ferroelectric, such as Zr doped HfO <sub...
Analog and RF mixed-signal cryogenic-CMOS circuits with ultrahigh gain-bandwidth product can address a range of applications such as interface between superconducting (SC) singleflux quantum (SFQ) logic cryo-dynamic random-access memory (DRAM), for sensing controlling qubits faster than their decoherence time at-scale processor.In this work, we evaluate performance 18 nm gate length (L G ) fully depleted silicon-on-insulator (FDSOI) NMOS PMOS from 300 to 5.5 K operating temperature.We...
Modern computers require an exponential increase in resources when solving computationally hard problems, motivating the need for alternative computing platform to solve such problems energy‐efficient manner. Vertex coloring, a nondeterministic polynomial time (NP‐hard) combinatorial optimization problem, is one problem. Herein, experimental demonstration of using cardiac cell‐based bio‐oscillator network coupling dynamics vertex coloring problem various scales graphs simple cell patterning...
Extended reality (XR) applications are machine learning (ML)-intensive, featuring deep neural networks (DNNs) with millions of weights, tightly latency-bound (10–20 ms end-to-end), and power-constrained (low tens mW average power). While ML performance efficiency can be achieved by introducing engines within low-power systems-on-chip (SoCs), system-level power for nontrivial DNNs depends strongly on the energy non-volatile memory (NVM) access network weights. This work introduces <italic...
As computational models inspired by the biological neural system, spiking networks (SNN) continue to demonstrate great potential in landscape of artificial intelligence, particularly tasks such as recognition, inference, and learning. While SNN focuses on achieving high-level intelligence individual creatures, Swarm Intelligence (SI) is another type bio-inspired that mimic collective swarms, i.e. bird flocks, fish school ant colonies. SI algorithms provide efficient practical solutions many...
Resistive switching devices—memristors—present a tunable, incremental behavior. Tuning their state accurately, repeatedly and in wide range, makes memristors well-suited for multi-level (ML) resistive memory cells analog computing applications. In this brief, the tuning approach based on memristor-resistor voltage divider (VD) is validated experimentally using commercial from <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Knowm Inc</i> ....
Ferroelectrics (FEs) are known to decompose into multidomain structures when combined with dielectric (DE) layers. According current understanding, it is thought that under such conditions, ferroelectric negative capacitance (NC) cannot be stabilized, and the operation of device would hysteretic. Here, we report, for first time, a tight control non-uniformity matching conditions nominally achieved, hysteresis-free can achieved in multi-domain FE-DE structure. We capture inhomogeneous...
Current rate of data generation and the need for real‐time analytics can benefit from new computational approaches where computation proceeds in a massively parallel way while being scalable energy efficient. Biological systems arising interaction living cells provide such pathways sustainable computing. designs biocomputing leveraging information processing units cells, as DNA, gene, or protein circuitries, are inherently slow (hours to days speed) and, therefore, primarily considered...
The maximum exploitation of the favorable properties and analog nature memristor technology in future nonvolatile resistive memories, requires accurate multi-level programming. In this direction, we explore voltage divider (VD) approach for highly controllable multi-state SET tuning. We present theoretical basis operation, main advantages weaknesses. finally propose an improved closed-loop VD scheme to tackle variability effect achieve <1% tuning precision, on average 3x faster than another...
The paradigm of biologically-inspired computing endows the components a neural network with dynamical functionality, such as self-oscillations, and harnesses emergent physical phenomena like synchronization, to learn classify complex temporal patterns. In this work, we exploit synchronization dynamics ultra-compact, low power Vanadium dioxide (VO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> ) based insulator-to-metal phase-transition...
Associative memory based on ternary content addressable (TCAM) is typically used in a static inference mode where the update occasional. However, for update-frequent associative search applications (e.g., clustering), existing TCAM designs have fundamental gap between low-density, high write performance SRAM, and high-density, poor nonvolatile memories. In this work, we demonstrate: i) monolithic 3D thin film transistors (TFT) that can simultaneously achieve density excellent performance,...