A5056241158- Evolutionary Algorithms and Applications
- Advanced Memory and Neural Computing
- VLSI and FPGA Design Techniques
- Neural Networks and Reservoir Computing
- Neural Networks and Applications
- Advanced Multi-Objective Optimization Algorithms
- Metaheuristic Optimization Algorithms Research
- VLSI and Analog Circuit Testing
- Gene Regulatory Network Analysis
- Low-power high-performance VLSI design
- Neural dynamics and brain function
- Advanced biosensing and bioanalysis techniques
- Modular Robots and Swarm Intelligence
- Embedded Systems Design Techniques
- Interconnection Networks and Systems
- Analog and Mixed-Signal Circuit Design
- Quantum-Dot Cellular Automata
- Evolution and Genetic Dynamics
- Parallel Computing and Optimization Techniques
- Cellular Automata and Applications
- Radiation Effects in Electronics
- Advanced Neural Network Applications
- Photonic and Optical Devices
- Photorefractive and Nonlinear Optics
- Reinforcement Learning in Robotics
University of York
2016-2025
Heidelberg University
2004-2006
Technische Universität Berlin
2002-2004
Neural networks have revolutionized the area of artificial intelligence and introduced transformative applications to almost every scientific field industry. However, this success comes at a great price; energy requirements for training advanced models are unsustainable. One promising way address pressing issue is by developing low-energy neuromorphic hardware that directly supports algorithm's requirements. The intrinsic non-volatility, non-linearity, memory spintronic devices make them...
The Reservoir Computing (RC) framework states that any non-linear, input-driven dynamical system (the reservoir) exhibiting properties such as a fading memory and input separability can be trained to perform computational tasks. This broad inclusion of systems has led many new physical substrates for RC. Properties essential reservoirs compute are tuned through reconfiguration the substrate, change in virtual topology or morphology. As result, each substrate possesses unique `quality' --...
Abstract We explore the effect of connectivity and topology on dynamical behaviour Reservoir Computers. At present, considerable effort is taken to design hand-craft physical reservoir computers. Both structure complexity are often pivotal task performance, however, assessing their overall importance challenging. Using a recently developed framework, we evaluate compare freedom (referring quality) neural network structures, as an analogy for systems. The results quantify how affects...
Field programmable gate arrays (FPGAs) are widely used in applications where online reconfigurable signal processing is required. Speed and function density of FPGAs increasing as transistor sizes shrink to the nanoscale. As these transistors reduce size intrinsic variability becomes more a problem reliably create electronic designs according specification time consuming statistical simulations become necessary; even with accurate models simulation, fabrication yield will decrease every...
Abstract Surface-immobilization of molecules can have a profound influence on their structure, function and dynamics. Toehold-mediated strand displacement is often used in solution to drive synthetic nanomachines made from DNA, but the effects surface-immobilization mechanism kinetics this reaction not yet been fully elucidated. Here we show that surface-immobilized are significantly different those phase attribute intermolecular interactions within DNA layer. We demonstrate dynamics be...
Recent work has shown that computational substrates made from carbon nanotube/polymer mixtures can form trainable Reservoir Computers. This new reservoir computing platform uses computer based evolutionary algorithms to optimise a set of electrical control signals induce properties within the substrate. In training process, evolution decides value analogue (voltages) and location inputs outputs on substrate improve performance subsequently trained readout. Here, we evaluate search compared...
The Reservoir Computing (RC) framework is said to have the potential transfer onto any input-driven dynamical system, provided two properties are present: (i) a fading memory, and (ii) input separability. A typical reservoir consists of fixed network recurrently connected processing units; however recent hardware implementations shown reservoirs not ultimately bound by this architecture. Previously, we demonstrated how RC can be applied randomly-formed carbon nanotube composites solve...
Reservoir Computing is an Unconventional Computation model to perform computation on various different substrates, such as RNNs or physical materials. The method takes a "black-box" approach, training only the outputs of system it built on. As such, evaluating computational capacity these systems can be challenging. We review and critique evaluation methods used in field Computing. introduce categorisation benchmark tasks. multiple examples benchmarks from literature applied reservoir...
Field programmable gate arrays (FPGAs) are widely used in applications where on-line reconfigurable signal processing is required. Speed and function density of FPGAs increasing when shrinking transistor sizes to the nano-scale. Unfortunately, order reliably create electronic designs according specification time-consuming statistical simulations become necessary due effects intrinsic variability. This paper describes an adaptive, evolvable architecture that allows for correction optimisation...
Biological genomes have evolved over a period of millions years and comprise thousands genes, even for the simplest organisms. However, in nature, only 1-2% genes play an active role creating maintaining organism, while majority are evolutionary fossils. This raises question whether considerably larger number (partly redundant) required order to effectively build functional developmental system, which, final system fraction is latter function. paper investigates different approaches...
The design of a new biologically inspired artificial developmental system is described in this paper. In general, systems converge slower and are more computationally expensive than direct evolution. However, the performance trends development indicate that full benefit will arise with larger complex problems exhibit some sort regularity their structure: thus, aim to evolve electronic through modularity allowed by development. hope proposed adaptivity fault tolerance future. cell signalling...
A model for intrinsic artificial development is introduced in this paper. The proposed features a novel mechanism where growth emerges, rather than being triggered by single action. Different types of cell signalling ensure that breaking symmetries the norm an exception, and gene activity regulated on two layers: first, proteins are produced regulatory network (GRN). Second, through structural feedback second messenger molecules, which not directly expression, but sensor proteins, take...
This work is part of a project that aims to develop and operate integrated evolvable hardware systems using unconstrained evolution. Experiments are carried out on an platform featuring both combinatorial registered logic as well sequential feedback loops. In order be able accurately assess the transient output system at same time speed up evolution, new fitness evaluation methods introduced. These bitwise hierarchical adapted further developed specifically for implementation. It shown newly...
A novel approach to image compression using a gene regulatory network (GRN) based artificial developmental system (ADS) is introduced. The proposed algorithm exploits the fact that series of complex organisms (≡ states) can be represented via GRN description and indices steps in which they occur. Organisms are interpreted as tiles an at each step results (re-)construction during process. It shown GRNs suitable for achieve higher rates than JPEG when optimised particular image. also same has...
Autonomously fault-tolerant systems have received a renewed interest for the design of dependable computing with increasing requirements variety critical applications including deep space probes, satellites, reactor control systems, and Internet-of-Things health environment monitoring. Autonomous are based on hardware capable self-monitoring self-repair. In this context, paper investigates use fine-grained, partial dynamic reconfiguration FPGA achieving higher degree fault-tolerance lower...
This paper presents XL-STaGe, a cross-layer tool for traffic-inclusive directed acyclic graph generation and implementation. In contrast to other graph-generation tools which focus on high-level DAG models, XL-STaGe consists of set processes that generate the task-graphs as well detailed process model each node in graph. The is highly customizable, with many parameters can be tuned meet user's requirements control topology, connection density, degree parallelism duration task-graph....
To tackle the complexity of state-of-the-art electronic systems, silicon foundries continuously shrink technology nodes and design automation (EDA) vendors offer hierarchical flows to decompose systems into smaller blocks. However, such a staged methodology consists various levels abstraction, where margins will be accumulated result in degradation overall quality. This limits full use capabilities both process EDA tools. In this work, study drive granularity standard cells is performed an...