Center for Brain Inspired Computing

We focus on model–hardware co-design for the deployment and integration of emerging AI technologies, such as physical neural networks and spiking neural networks, within digital processing systems. We develop algorithms and computer architectures that leverage the unique characteristics of brain-inspired compute elements within heterogeneous hardware platforms to enable next-generation embedded AI. 

Main PI: dr.ir. N. Alachiotis (Nikolaos)

We develop algorithms for high-performance computing that deliver accuracy at reduced cost. Two main principles underpin our research into new algorithms: (i) structure preservation and (ii) model order reduction. We combine deterministic and stochastic modelling and incorporate data assimilation to derive virtual twins of technological systems found in neuromorphic computing.

Main PI: Prof.dr.ir. B.J. Geurts (Bernard)

We develop brain-inspired nanoelectronic hardware where physical matter performs computation. Our group has pioneered nanoscale reconfigurable nonlinear computing devices (disordered nanoparticle networks, silicon reconfigurable nonlinear-processing units, RNPUs) that process signals directly in hardware, opening a pathway to ultra-energy-efficient inference. Our research spans fundamental device physics to scalable chips for edge AI and novel computing paradigms.

Main PI: prof.dr.ir. W.G. van der Wiel (Wilfred)

A central focus of my lab is to understand how simple physical principles can give rise to life-like behaviors such as autonomous motion, dynamic self-organization, and adaptability. To address this challenge, my lab pursues TWO complementary research lines: synthetic active matter and active lipid vesicles. Both are rooted in nonequilibrium physics and driven by the ambition to design reconfigurable and adaptive, bioinspired materials.

Main PI: dr. H.R. Vutukuri PhD (Hanumantha Rao)

Chemical Reaction Netorks (CRNs) networks are ubiquitous in living systems (genetic, protein, and metabolic networks), enabling storage, transmission, and information processing through nonlinear biochemical interactions. While this complexity offers a large potential for neuromorphic computing it is unknown how exactly this chemistry can used for an alternative energy-efficient computing technology. We pioneered how CRNs can enable memory, molecular information processing, and—more generally—computing. Building on this foundation, Wong and his team aim to design an experimental platform for Chemical Information Processing (CHIP), promising a fundamentally new type of artificial intelligence that is governed by networks of chemical reactions.

Main PI: dr.ing. S.Y. Wong (Albert)

Our group focuses on materials and devices of which the electric properties are coupled to (local) state variables. We study the relation between nanoscopic material properties, tunable by growth, and the rich static and dynamic responses of our devices to input signals.

Main PI's: 
prof.dr.ir. J.W.M. Hilgenkamp (Hans)  
dr.ir. A.E.M. Smink (Sander)

We develop ferroelectric tunnel junctions as memristive devices for crossbar arrays in in-memory computing. For this purpose, reliable growth of nanometer-thick ferroelectric films is essential. We study BaTiO₃, focusing on electrode interface and defect effects, and Sm_0.2Bi_1.8O₃, a recently identified ultrathin ferroelectric, using pulsed laser deposition.

Main PI's:
prof.dr.ir. G. Koster (Gertjan)
prof.dr.ing. A.J.H.M. Rijnders (Guus)

Neuromorphic Sensing and processing

• Low Power AI processors
• Hardware aware algorithm design

I extensively use digital hardware design tools and gate-level simulations to co-optimize hardware architectures with neural network algorithms. Some of my technical experiences:

• Design of programmable neuromorphic processors for embedded AI applications
• On-device learning algorithms and hardware accelerators
• Bio-inspired vision processing
• Benchmarking and comparison of various algorithm optimizations in hardware

Main PI: A. Yousefzadeh PhD (Amir)

We harness the natural dynamics of molecular materials to create intelligent hardware for energy efficient information processing. The dynamical properties enable event-driven spike-based neural networks that mimic computing biological systems like our Brains. This approach simplifies system architectures, is fault tolerant, and operates at low drive voltages greatly reducing power consumption.

Main PIs: 
Dr. Ir. Sissi de Beer
Dr. Ivana Lin
Prof. Dr. Christian Nijhuis