In order to develop and evaluate the extent to which itching affects a person's daily life, it is useful to develop automated means to recognise the action of scratching. We present an investigation of sensors and algorithms to realise a wearable scratch detection device. We collected a dataset, where each user wore 4 inertial measurement unit (IMU) sensors and one electric potential sensor (EPS). Data were collected from nine users, where each user followed a 40-min protocol, which involved scratching different parts of head, shoulder, and leg, as well as other activities such as walking, drinking water, brushing teeth, and typing to a computer. The dataset contained 813 scratching instances and 5 h 15 min of recorded data. We investigated the trade-offs between the number of devices worn (comfort) and accuracy. We trained the k-NN and random forest algorithms by using between 1 and 5 features per channel. We concluded that a scratch could be detected with 80.7% accuracy by using the random forest algorithm on hand coordinates, which required four devices. However, an f1 score of 70% could be achieved with k-NN with IMU and EPS data, which only required one device. Moreover, the fusion of IMU data with EPS data improved the accuracy and reduced the deviation between the folds. This expanded the state-of-the-art method by opening up new trade-offs between accuracy and comfort for future research

Conceptors are a recent development in the field of reservoir computing; they can be used to influence the dynamics of recurrent neural networks (RNNs), enabling generation of arbitrary patterns based on training data. Conceptors allow interpolation and extrapolation between patterns, and also provide a system of boolean logic for combining patterns together. Generation and manipulation of arbitrary patterns using conceptors has significant potential as a sound synthesis method for applications in computer music but has yet to be explored. Conceptors are untested with the generation of multi-timbre audio patterns, and little testing has been done on scalability to longer patterns required for audio. A novel method of sound synthesis based on conceptors is introduced. Conceptular Synthesis is based on granular synthesis; sets of conceptors are trained to recall varying patterns from a single RNN, then a runtime mechanism switches between them, generating short patterns which are recombined into a longer sound. The quality of sound resynthesis using this technique is experimentally evaluated. Conceptor models are shown to resynthesise audio with a comparable quality to a close equivalent technique using echo state networks with stored patterns and output feedback. Conceptor models are also shown to excel in their malleability and potential for creative sound manipulation, in comparison to echo state network models which tend to fail when the same manipulations are applied. Examples are given demonstrating creative sonic possibilities, by exploiting conceptor pattern morphing, boolean conceptor logic and manipulation of RNN dynamics. Limitations of conceptor models are revealed with regards to reproduction quality, and pragmatic limitations are also shown, where rises in computation and memory requirements preclude the use of these models for training with longer sound samples. The techniques presented here represent an initial exploration of the sound synthesis potential of conceptors, demonstrating possible creative applications in sound design; future possibilities and research questions are outlined.

Echo State Networks (ESNs), a form of recurrent neural network developed in the field of Reservoir Computing, show significant potential for use as a tool in the design of mappings for digital musical instruments. They have, however, seldom been used in this area, so this paper explores their possible uses. This project contributes a new open source library, which was developed to allow ESNs to run in the Pure Data dataflow environment. Several use cases were explored, focusing on addressing current issues in mapping research. ESNs were found to work successfully in scenarios of pattern classification, multiparametric control, explorative mapping and the design of nonlinearities and uncontrol. \emph{Un-trained} behaviours are proposed, as augmentations to the conventional reservoir system that allow the player to introduce potentially interesting non-linearities and uncontrol into the reservoir. Interactive evolution style controls are proposed as strategies to help design these behaviours, which are otherwise dependent on arbitrary parameters. A study on sound classification shows that ESNs can reliably differentiate between two drum sounds, and also generalise to other similar input. Following evaluation of the use cases, heuristics are proposed to aid the use of ESNs in computer music scenarios.

In this paper we summarize the contributions of participants to the Sussex-Huawei Transportation-Locomotion (SHL) Recognition Challenge organized at the HASCA Workshop of UbiComp 2018. The SHL challenge is a machine learning and data science competition, which aims to recognize eight transportation activities (Still, Walk, Run, Bike, Bus, Car, Train, Subway) from the inertial and pressure sensor data of a smartphone. We introduce the dataset used in the challenge and the protocol for the competition. We present a meta-analysis of the contributions from 19 submissions, their approaches, the software tools used, computational cost and the achieved results. Overall, two entries achieved F1 scores above 90%, eight with F1 scores between 80% and 90%, and nine between 50% and 80%.

In this paper we, as part of the Sussex-Huawei Locomotion-Transportation (SHL) Recognition Challenge organizing team, present reference recognition performance obtained by applying various classical and deep-learning classifiers to the testing dataset. We aim to recognize eight modes of transportation (Still, Walk, Run, Bike, Bus, Car, Train, Subway) from smartphone inertial sensors: accelerometer, gyroscope and magnetometer. The classical classifiers include naive Bayesian, decision tree, random forest, K-nearest neighbour and support vector machine, while the deep-learning classifiers include fully-connected and convolutional deep neural networks. We feed different types of input to the classifier, including hand-crafted features, raw sensor data in the time domain, and in the frequency domain. We employ a post-processing scheme to improve the recognition performance. Results show that convolutional neural network operating on frequency domain raw data achieves the best performance among all the classifiers.

As augmented reality (AR) quickly evolves with new technological practice, there is a growing need to question and reevaluate its potential as a medium for creative expression. The authors discuss AR within computational art, framed within AR as a medium, AR aesthetics and applications. The Forum for Augmented Reality Immersive Instruments (ARImI), a two-day event on AR, highlights both possibilities and fundamental concerns for continuing artworks in this field, including visual bias, sensory modalities, interactivity and performativity. The authors offer a new AR definition as real-time computationally mediated perception.

10000 neural oscillators conspire to make a bad approximation of a one second video loop, while their internal networks are stretched and twisted by their own sound and their neighbours.

This is an improvised performance, where a video is resynthesised by an ensemble of neural conceptor networks, each of which has learnt (with varying degrees of success) to approximate the behaviour of an individual pixel in the original. The ghostly result is manipulated live, to send this mass of miniature feedback systems into collective overdrive.

We demonstrate a method to encode complex human gestures acquired from inertial sensors for activity recognition. Gestures are encoded as a stream of symbols which represent the change in orientation and displacement of the body limbs over time.
The first novelty of this encoding is to enable the reuse previously developed single-channel template matching algorithms also when multiple sensors are used simultaneously.
The second novelty is to encode changes in orientation of limbs which is important in some activities, such as sport analytics.
We demonstrate the method using our custom inertial platform, BlueSense. Using a set of five BlueSense nodes, we implemented a motion tracking system that displays a 3D human model and shows in real-time the corresponding movement encoding.

During active behaviours like running, swimming, whisking or sniffing, motor actions shape sensory input and sensory percepts guide future motor commands. Ongoing cycles of sensory and motor processing constitute a closed-loop feedback system which is central to motor control and, it has been argued, for perceptual processes. This closed-loop feedback is mediated by brainwide neural circuits but how the presence of feedback signals impacts on the dynamics and function of neurons is not well understood. Here we present a simple theory suggesting that closed-loop feedback between the brain/body/environment can modulate neural gain and, consequently, change endogenous neural fluctuations and responses to sensory input. We support this theory with modeling and data analysis in two vertebrate systems. First, in a model of rodent whisking we show that negative feedback mediated by whisking vibrissa can suppress coherent neural fluctuations and neural responses to sensory input in the barrel cortex. We argue this suppression provides an appealing account of a brain state transition (a marked change in global brain activity) coincident with the onset of whisking in rodents. Moreover, this mechanism suggests a novel signal detection mechanism that selectively accentuates active, rather than passive, whisker touch signals. This mechanism is consistent with a predictive coding strategy that is sensitive to the consequences of motor actions rather than the difference between the predicted and actual sensory input. We further support the theory by re-analysing previously published two-photon data recorded in zebrafish larvae performing closed-loop optomotor behaviour in a virtual swim simulator. We show, as predicted by this theory, that the degree to which each cell contributes in linking sensory and motor signals well explains how much its neural fluctuations are suppressed by closed-loop optomotor behaviour. More generally we argue that our results demonstrate the dependence of neural fluctuations, across the brain, on closed-loop brain/body/environment interactions strongly supporting the idea that brain function cannot be fully understood through open-loop approaches alone.

To increase fitness for survival, organisms not only passively react to environmental changes but also actively predict future events to prepare for potential hazards within their environment. Accumulating evidence indicates that the human brain is a remarkable predictive machine which constantly models causal relationships and predicts future events. This ‘predictive processing’ framework, a prediction-based form of Bayesian inference, states that the brain continuously generates and updates predictions about incoming sensory signals. This framework has been showing notable explanatory power in understanding the mechanisms behind both human behaviour and neurophysiological data and elegantly specifies the underlying computational principles of the neural system. However, even though predictive processing has the potential to provide a unified theory of the brain (Karl Friston, 2010), we still have a limited understanding about fundamental aspects of this model, such as how it deals with different types of information, learns statistical regularities and perhaps most fundamentally of all what its relationship to conscious experience is. This thesis aims to investigate the major gaps in our current understanding of the predictive processing framework via a series of studies. Study 1 investigated the fundamental relationship between unconscious statistical inference reflected by predictive processing and conscious access. It demonstrated that predictions that are in line with sensory evidence accelerate conscious access. Study 2 investigated how low level information within the sensory hierarchy is dealt with by predictive processing and regularity learning mechanisms through “perceptual echo” in which the cross-correlation between a sequence of randomly fluctuating luminance values and occipital electrophysiological (EEG) signals exhibits a long-lasting periodic (~100ms cycle) reverberation of the input stimulus. This study identified a new form of regularity learning and the results demonstrate that the perceptual echo may reflect an iterative learning process, governed by predictive processing. Study 3 investigated how supra-modal predictive processing is capable of
learning regularities of temporal duration and also temporal predictions about future events. This study revealed a supramodal temporal prediction mechanism which processes auditory and visual temporal information and integrates information from the duration and rhythmic structures of events. Together these studies provide a global picture of predictive processing and regularity learning across differing types of predictive information.

Presented for Boundary Work II was a collection of key works from Micheál O'Connell aka Mocksim's artistic practice which incorporates the use of digital tools and performance. This included short looping films, live art and 'stupid' applications of science, photography, technology. The show included 'Now Man', an activity in which participants are challenged to respond to the movements of a rapidly spinning camera, described as a game of "digital narcissism, heady sadism and running about". Also on display will be Contra-Invention, nominated for the Deutsche Börse Photography Prize 2012, an exhibition of photographs taken by traffic wardens of illegally parked cars in one British town.

We present a novel unsupervised approach, UnADevs, for discovering activity clusters corresponding to periodic and stationary activities in streaming sensor data. Such activities usually last for some time, which is exploited by our method; it includes mechanisms to regulate sensitivity to brief outliers and can discover multiple clusters overlapping in time to better deal with deviations from nominal behaviour. The method was evaluated on two activity datasets containing large number of activities (14 and 33 respectively) against online agglomerative clustering and DBSCAN. In a multi-criteria evaluation, our approach achieved significantly better performance on majority of the measures, with the advantages that: (i) it does not require to specify the number of clusters beforehand (it is open ended); (ii) it is online and can find clusters in real time; (iii) it has constant time complexity; (iv) and it is memory efficient as it does not keep the data samples in memory. Overall, it has managed to discover 616 of the total 717 activities. Because it discovers clusters of activities in real time, it is ideal to work alongside an active learning system.

Neuromorphic computing employs models of neuronal circuits to solve computing problems. Neuromorphic hardware systems are now becoming more widely available and “neuromorphic algorithms” are being developed. As they are maturing toward deployment in general research environments, it becomes important to assess and compare them in the context of the applications they are meant to solve. This should encompass not just task performance, but also ease of implementation, speed of processing, scalability, and power efficiency. Here, we report our practical experience of implementing a bio-inspired, spiking network for multivariate classification on three different platforms: the hybrid digital/analog Spikey system, the digital spike-based SpiNNaker system, and GeNN, a meta-compiler for parallel GPU hardware. We assess performance using a standard hand-written digit classification task. We found that whilst a different implementation approach was required for each platform, classification performances remained in line. This suggests that all three implementations were able to exercise the model’s ability to solve the task rather than exposing inherent platform limits, although differences emerged when capacity was approached. With respect to execution speed and power consumption, we found that for each platform a large fraction of the computing time was spent outside of the neuromorphic device, on the host machine. Time was spent in a range of combinations of preparing the model, encoding suitable input spiking data, shifting data, and decoding spike-encoded results. This is also where a large proportion of the total power was consumed, most markedly for the SpiNNaker and Spikey systems. We conclude that the simulation efficiency advantage of the assessed specialized hardware systems is easily lost in excessive host-device communication, or non-neuronal parts of the computation. These results emphasize the need to optimize the host-device communication architecture for scalability, maximum throughput, and minimum latency. Moreover, our results indicate that special attention should be paid to minimize host-device communication when designing and implementing networks for efficient neuromorphic computing.

At the same time as we are seeing ever increasing numbers of people actively using social networking sites, and growing evidence of increased participation in campaigning and digital activism, we are seeing a decline in democratic participation in the UK at both a national and local level. This thesis examines these two contrasting effects within the context of Local Government in the UK and explores what the impact might be at the neighbourhood level. The work discusses the influence of place based online activity on democratic decision-making Local Government and the ways in which traditional processes of decision-making, democratic participation and community engagement practice may need to change to reflect the upward pressure that is being exerted by citizen use of new technologies and adjust the way in which Local Government facilitates citizen participation in decision-making. The work develops the concept of Digital civic space as an alternative to eParticipation platforms and discusses how such spaces are being used to connect online activity with democratic processes at present and how present experience may be used to inform future developments. Employing an Action Research method, the research analyses three projects in order to examine the nature of the pre-existing participation online and the impact of creating online civic spaces to connect the participants both to each other and to local decision-makers. Design criteria are proposed which describe the necessary qualities of public-ness, openness, co-production, definition of place and identity and the thesis reaches conclusions as to how these criteria might better connect local resident with the democratic decision-making processes for their communities.

A mutant of the high fidelity family-B DNA polymerase from the archaeon Thermococcus gorgonarius (Tgo-Pol), able to replicate past DNA lesions, is described. Gain of function requires replacement of the three amino acid loop region in the fingers domain of Tgo-Pol with a longer version, found naturally in eukaryotic Pol zeta (a family-B translesion synthesis polymerase). Inactivation of the 3'–5' proofreading exonuclease activity is also necessary. The resulting Tgo-Pol Z1 variant is proficient at initiating replication from base mismatches and can read through damaged bases, such as abasic sites and thymine photo-dimers. Tgo-Pol Z1 is also proficient at extending from primers that terminate opposite aberrant bases. The fidelity of Tgo-Pol Z1 is reduced, with amarked tendency tomake changes at G:C base pairs. Together, these results suggest that the loop region of the fingers domain may play a critical role in determining whether a family-B enzyme falls into the accurate genome-replicating category or is an errorprone translesion synthesis polymerase. Tgo-Pol Z1 may also be useful for amplification of damaged DNA.

Induction of chemokine gene expression from peripheral blood mononuclear cells (PBMCs) stimulated by proinflammatory cytokines plays an important role in both wound repair and response to infectious agents. In the present study, we show that the proinflammatory cytokine interleukin-6 (IL-6) potently induced mRNA expression and secretion of the CC chemokine monocyte chemotactic protein 1 (MCP-1) in PBMCs. In addition, because human immunodeficiency virus (HIV) infection in vivo and in vitro has been shown to dysregulate the production of and/or the response to cytokines, PBMCs from both healthy uninfected and HIV-infected individuals were studied for their constitutive and IL-6-induced expression of MCP-1. No substantial differences were observed between the two groups of individuals. In addition, IL-6 upregulated MCP-1 expression in the promonocytic cell line U937 and in its chronically HIV-infected counterpart, U1. In these cell lines, IL-6 selectively induced MCP-1 and not other chemokines, including regulated upon activation normal T cells expressed and secreted (RANTES), macrophage inflammatory protein-1alpha (MIP-1alpha), MIP-1beta, and IL-8. IL-6 induction of MCP-1 was partially inhibited by hydrocortisone in U1 cells. Thus, IL-6 activates PBMCs to secrete MCP-1, a CC chemokine pivotal for monocyte recruitment in tissue and organs in which important inflammatory events occur.

The emergence of a unified cognitive behaviour relies on the coordination of specialized components that distribute across a ‘brain’, body and environment. Although a general dynamical mechanism involved in agent–environment integration is still largely unknown for behavioural robustness, discussions here are focussed on one of the most plausible candidate: the formation of distributed mechanisms working in transient during agent–environment coupling. This article provides discussions on this sort of coordination based on a mobile object-tracking task with situated, embodied and minimal agents, and tests for robust yet adaptive behaviour. The proposed scenario provides examples of behavioural mechanisms that counterbalance the functional organization of internal control activity and agents’ situatedness to enable the evolution of a two-agent interaction task. Discussions in this article suggest that future studies of distributed cognition should take into account that there are at least two possible modes of interpreting distributed mechanisms and that these have a qualitatively different effect on behavioural robustness.

Theoretical discussions and computational models of bio-inspired embodied and situated agents are introduced in this article capturing in simplified form the dynamical essence of robust, yet adaptive behavior. This article analyzes the general problem of how the dynamical coupling between internal control (brain), body and environment is used in the generation of specific behaviors. Based on the Evolutionary Robotics (ER) paradigm, four computational models are described to support discussions including descriptions on performance after a series of structural, sensorimotor or mutational perturbations, or are developed in the absence of them. Experimental results suggest that ‘dynamic determinacy’ – i.e. the continuous presence of a unique dynamical attractor that must be chased during functional behaviors – is a common dynamic phenomenon in the analyzed robust and adaptive agents. These agents show dynamical states that are definitely and unequivocally characterized via transient dynamics toward a unique, yet moving attractor at neural level for coherent actions. This determinacy emerges as a control strategy rooted on behavioral couplings and relies on mechanisms that are distributed on brain, body and environment. Different ways to induce further distribution of behavioral mechanisms are also discussed in this paper from a bio-inspired ER perspective.

This article investigates the emergence of robust behaviour in agents with dynamically limited controllers (monostable agents), and compares their performance to less limited ones (bistable agents). ‘Dynamically limited’ here refers to a reduced quantity of steady states that an agent controller exhibits when it does not receive stimulus from the environment. Agents are evolved for categorical perception, a minimal cognitive task, and must correlate approaching or avoiding movements based on (two) different types of objects. Results indicate a significant tendency to better behaviouralrobustness by monostable in contrast to bistable agents in the presence of sensorimotor, mutational, and structural perturbations. Discussions here focus on a further dependence to coupled dynamics by the former agents to explain such a tendency.