Source author record

Stuart Watt

Stuart Watt appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

cond-mat.dis-nn cond-mat.mtrl-sci Genomics Neural and Evolutionary Computing physics.app-ph Quantitative Methods

Catalog footprint

What is connected

4works

6topics

3close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2020arXiv

Spoken digit classification using a spin-wave delay-line active-ring reservoir computing

As a test of general applicability, we use the recently proposed spin-wave delay line active-ring reservoir computer to perform the spoken digit recognition task. On this, classification accuracies of up to 93% are achieved. The tested device prototype employs improved spin wave transducers (antennas). Therefore, in addition, we also let the computer complete the short-term memory (STM) task and the parity check (PC) tasks, because the fading memory and nonlinearity are essential to reservoir computing performance. The resulting STM and PC capacities reach maximum values of 4.77 and 1.47 respectively.

preprint2019arXiv

Manipulation of Inverse Spin Hall Effect in Palladium by Absorption of Hydrogen Gas

The spintronic properties of a palladium thin film have been investigated in the presence of hydrogen gas in cobalt/palladium bilayers. Measurements of the inverse spin Hall Effect (ISHE) using cavity ferromagnetic resonance allow estimations of the spin Hall conductivity and spin diffusion length in both nitrogen and hydrogen gas atmospheres. Unwanted spin rectification effects are removed using a simple method of inverting the spin current direction with respect to the measurement setup. Absorption of hydrogen gas in the Pd layer at just 3% concentration results in a reduced ISHE voltage amplitude and bilayer resistance. Fitting the ISHE voltage against the Pd layer thickness demonstrates that the spin diffusion length decreases by 23% in the presence of a hydrogen gas. On the other hand, the results indicate that there is no significant change in the spin Hall conductivity of Pd due to hydrogen absorption.

preprint2019arXiv

Reservoir computing using a spin-wave delay line active ring resonator

The authors demonstrate the use of a propagating spin waves for implementing a reservoir computing architecture. The proposed concept utilises an active ring resonator comprising a magnetic thin film delay line integrated into a feedback loop. These systems exhibit strong nonlinearity and delayed response behaviour, two important properties required for an effective reservoir computing implementation. In a simple design, we exploit the nonlinear damping of spin waves at different feedback gains to inject input data into the active ring resonator and use a microwave diode to read out the amplitude of the spin waves circulating in the ring. We employ two baseline tasks, namely the short term memory and parity check tasks, to evaluate the suitability of this architecture for processing time series data.

preprint2015arXiv

Integrating alignment-based and alignment-free sequence similarity measures for biological sequence classification

Alignment-based sequence similarity searches, while accurate for some type of sequences, can produce incorrect results when used on more divergent but functionally related sequences that have undergone the sequence rearrangements observed in many bacterial and viral genomes. Here, we propose a classification model that exploits the complementary nature of alignment-based and alignment-free similarity measures with the aim to improve the accuracy with which DNA and protein sequences are characterized. Our model classifies sequences using a combined sequence similarity score calculated by adaptively weighting the contribution of different sequence similarity measures. Weights are determined independently for each sequence in the test set and reflect the discriminatory ability of individual similarity measures in the training set. Since the similarity between some sequences is determined more accurately with one type of measure rather than another, our classifier allows different sets of weights to be associated with different sequences. Using five different similarity measures we show that our model significantly improves the classification accuracy over the current composition and alignment based models, when predicting the taxonomic lineage for both short viral sequence fragments and complete viral sequences. We also show that our model can be used effectively for the classification of reads from a real metagenome dataset as well as protein sequences.