Search

We demonstrate the use of a convolutional neural network to perform neutron-gamma pulse shape discrimination, where the only inputs to the network are the raw digitised SiPM signals from a dual scintillator detector element made of 6LiF:ZnS(Ag) scintillator and PVT plastic. A realistic labelled dataset was created to train the network by exposing the detector to an AmBe source, and a data-driven method utilsing a separate PMT was used to assign labels to the recorded signals. This approach is compared to the charge integration and continuous wavelet transform methods and is found to provide superior levels of discrimination, achieving an AUC of 0.995 +/- 0.003. We find that the neural network is capable of extracting interpretable features directly from the raw data. In addition, by visualising the high-dimensional representations of the network with the t-SNE algorithm, we discover that not only is this method robust to minor mislabeling of the training dataset but that it is possible to identify an underlying substructure within the signals that goes beyond the original labelling. This technique could be utilised to explore and cluster complex, raw detector data in a novel way that may reveal more insights than standard analysis methods., 13 pages, 8 figures, submitted to Nucl. Instr. Meth. A

In this work, the capability of restricted Boltzmann machines (RBMs) to find solutions for the Kitaev honeycomb model with periodic boundary conditions is investigated. The measured groundstate (GS) energy of the system is compared and, for small lattice sizes (e.g. $3 \times 3$ with $18$ spinors), shown to agree with the analytically derived value of the energy up to a deviation of $0.09\%$. Moreover, the wave-functions we find have $99.89\%$ overlap with the exact ground state wave-functions. Furthermore, the possibility of realizing anyons in the RBM is discussed and an algorithm is given to build these anyonic excitations and braid them for possible future applications in quantum computation. Using the correspondence between topological field theories in (2+1)d and 2d CFTs, we propose an identification between our RBM states with the Moore-Read state and conformal blocks of the $2$d Ising model., Comment: (15+5) pages, 11 figures, 5 tables: Minor journal revisions and additions. + Journal ref

Funder: Royal Society; doi: http://dx.doi.org/10.13039/501100000288, Funder: Trinity College, University of Cambridge; doi: http://dx.doi.org/10.13039/501100000727, Funder: John Fell Fund, University of Oxford; doi: http://dx.doi.org/10.13039/501100004789, We use a deep neural network (NN) to detect and place region-of-interest (ROI) boxes around ultracold atom clouds in absorption and fluorescence images—with the ability to identify and bound multiple clouds within a single image. The NN also outputs segmentation masks that identify the size, shape and orientation of each cloud from which we extract the clouds’ Gaussian parameters. This allows 2D Gaussian fits to be reliably seeded thereby enabling fully automatic image processing. The method developed performs significantly better than a more conventional method based on a standardized image analysis library (Scikit-image) both for identifying ROI and extracting Gaussian parameters.

Calculating polarizabilities of large clusters with first-principles techniques is challenging because of the unfavorable scaling of computational cost with cluster size. To address this challenge, we demonstrate that polarizabilities of large hydrogenated silicon clusters containing thousands of atoms can be efficiently calculated with machine learning methods. Specifically, we construct machine learning models based on the smooth overlap of atomic positions (SOAP) descriptor and train the models using a database of calculated random-phase approximation polarizabilities for clusters containing up to 110 silicon atoms. We first demonstrate the ability of the machine learning models to fit the data and then assess their ability to predict cluster polarizabilities using k-fold cross validation. Finally, we study the machine learning predictions for clusters that are too large for explicit first-principles calculations and find that they accurately describe the dependence of the polarizabilities on the ratio of hydrogen to silicon atoms and also predict a bulk limit that is in good agreement with previous studies.

Funder: Trinity College, University of Cambridge; doi: http://dx.doi.org/10.13039/501100000727, Selecting the most relevant features and samples out of a large set of candidates is a task that occurs very often in the context of automated data analysis, where it improves the computational performance and often the transferability of a model. Here we focus on two popular subselection schemes applied to this end: CUR decomposition, derived from a low-rank approximation of the feature matrix, and farthest point sampling (FPS), which relies on the iterative identification of the most diverse samples and discriminating features. We modify these unsupervised approaches, incorporating a supervised component following the same spirit as the principal covariates (PCov) regression method. We show how this results in selections that perform better in supervised tasks, demonstrating with models of increasing complexity, from ridge regression to kernel ridge regression and finally feed-forward neural networks. We also present adjustments to minimise the impact of any subselection when performing unsupervised tasks. We demonstrate the significant improvements associated with PCov-CUR and PCov-FPS selections for applications to chemistry and materials science, typically reducing by a factor of two the number of features and samples required to achieve a given level of regression accuracy.

Significance: Acoustically detecting the rich optical absorption contrast in biological tissues, photoacoustic tomography (PAT) seamlessly bridges the functional and molecular sensitivity of optical excitation with the deep penetration and high scalability of ultrasound detection. As a result of continuous technological innovations and commercial development, PAT has been playing an increasingly important role in life sciences and patient care, including functional brain imaging, smart drug delivery, early cancer diagnosis, and interventional therapy guidance. Aim: Built on our 2016 tutorial article that focused on the principles and implementations of PAT, this perspective aims to provide an update on the exciting technical advances in PAT. Approach: This perspective focuses on the recent PAT innovations in volumetric deep-tissue imaging, high-speed wide-field microscopic imaging, high-sensitivity optical ultrasound detection, and machine-learning enhanced image reconstruction and data processing. Representative applications are introduced to demonstrate these enabling technical breakthroughs in biomedical research. Conclusions: We conclude the perspective by discussing the future development of PAT technologies.

We introduce and explore an approach for constructing force fields for small molecules, which combines intuitive low body order empirical force field terms with the concepts of data driven statistical fits of recent machine learned potentials. We bring these two key ideas together to bridge the gap between established empirical force fields that have a high degree of transferability on the one hand, and the machine learned potentials that are systematically improvable and can converge to very high accuracy, on the other. Our framework extends the atomic permutationally invariant polynomials (aPIP) developed for elemental materials in (2019 Mach. Learn.: Sci. Technol. 1 015004) to molecular systems. The body order decomposition allows us to keep the dimensionality of each term low, while the use of an iterative fitting scheme as well as regularisation procedures improve the extrapolation outside the training set. We investigate aPIP force fields with up to generalised 4-body terms, and examine the performance on a set of small organic molecules. We achieve a high level of accuracy when fitting individual molecules, comparable to those of the many-body machine learned force fields. Fitted to a combined training set of short linear alkanes, the accuracy of the aPIP force field still significantly exceeds what can be expected from classical empirical force fields, while retaining reasonable transferability to both configurations far from the training set and to new molecules.

Due to their portability, versatility for supporting multiple assay formats, and potential for resulting in low-cost assays, paper-based analytical devices (PADs) are an increasingly popular format as a platform for the development of point-of-care tests. However, very few PADs have been translated successfully to their intended environments outside of academic settings. Often overlooked as a factor that inhibits translation, usability is a vital characteristic of any successful point-of-care test. Recent advancements in PAD design have demonstrated improved usability by simplifying various aspects of user operation, including sample collection, sample processing, device operation, detection, and readout/interpretation. Field testing at various stages of device design can offer critical feedback about device usability, especially when it involves the proposed end-user or other stakeholders. By highlighting advances in usability, we aim to encourage thoughtful and rigorous design at the academic prototyping stage to address one outstanding hurdle that limits the number of PADs that make it from the benchtop to the point-of-care.

Bayesian optimisation is a sample-efficient search methodology that holds great promise for accelerating drug and materials discovery programs. A frequently-overlooked modelling consideration in Bayesian optimisation strategies however, is the representation of heteroscedastic aleatoric uncertainty. In many practical applications it is desirable to identify inputs with low aleatoric noise, an example of which might be a material composition which consistently displays robust properties in response to a noisy fabrication process. In this paper, we propose a heteroscedastic Bayesian optimisation scheme capable of representing and minimising aleatoric noise across the input space. Our scheme employs a heteroscedastic Gaussian process (GP) surrogate model in conjunction with two straightforward adaptations of existing acquisition functions. First, we extend the augmented expected improvement (AEI) heuristic to the heteroscedastic setting and second, we introduce the aleatoric noise-penalised expected improvement (ANPEI) heuristic. Both methodologies are capable of penalising aleatoric noise in the suggestions and yield improved performance relative to homoscedastic Bayesian optimisation and random sampling on toy problems as well as on two real-world scientific datasets. Code is available at: \url{https://github.com/Ryan-Rhys/Heteroscedastic-BO}, Published in Machine Learning: Science and Technology 2021 (https://iopscience.iop.org/article/10.1088/2632-2153/ac298c) Earlier version accepted to the 2019 NeurIPS Workshop on Safety and Robustness in Decision Making

Inconsistencies among testing methods and results in previous research prompted this study that builds upon a systematic usability testing research framework to better understand how interface medium influences users' abilities to detect usability flaws in applications. Interface medium was tested to identify its effects on users' perceptions of usability and abilities to detect usability problems and severe usability problems. Results indicated that medium has no effect on users' abilities to detect usability problems or perceptions of usability. However, results did indicate an interaction between the medium and the tested application in which users were able to identify significantly more usability problems on a higher fidelity medium using a particular application. Results also indicated that as users' perceptions of an application's usability increases, the users are less able to detect usability problems in that application. Usability testing should begin early in the design process, even if low fidelity mediums will be used.

Four experiments were conducted to test whether recent developments in display technology would suffice to eliminate the well-known disadvantages in reading from screen as compared with paper. Proofreading speed and performance were equal for a TFT-LCD and a paper display, but there were more symptoms of eyestrain in the screen condition accompanied by a strong preference for paper (Experiment 1). These results were replicated using a longer reading duration (Experiment 2). Additional experiments were conducted to test hypotheses about the reasons for the higher amount of eyestrain associated with reading from screen. Reduced screen luminance did not change the pattern of results (Experiment 3), but positioning both displays in equal inclination angles eliminated the differences in eyestrain symptoms and increased proofreading speed in the screen condition (Experiment 4). A paper-like positioning of TFT-LCDs seems to enable unimpaired reading without evidence of increased physical strain. Practitioner Summary: Given the developments in screen technology, a re-assessment of the differences in proofreading speed and performance, well-being, and preference between computer screen and paper was conducted. State-of-the-art TFT-LCDs enable unimpaired reading, but a book-like positioning of screens seems necessary to minimise eyestrain symptoms.

This study examines whether the Internet-based questionnaire is psychometrically equivalent to the paper-based questionnaire. A random sample of 2,400 teachers in Taiwan was divided into experimental and control groups. The experimental group was invited to complete the electronic form of the Chinese version of Center for Epidemiologic Studies Depression Scale (CES-D) placed on the Internet, whereas the control group was invited to complete the paper-based CES-D, which they received by mail. The multisample invariance approach, derived from structural equation modeling (SEM), was applied to analyze the collected data. The analytical results show that the two groups have equivalent factor structures in the CES-D. That is, the items in CES-D function equivalently in the two groups. Then the equality of latent mean test was performed. The latent means of "depressed mood," "positive affect," and "interpersonal problems" in CES-D are not significantly different between these two groups. However, the difference in the "somatic symptoms" latent means between these two groups is statistically significant at alpha = 0.01. But the Cohen's d statistics indicates that such differences in latent means do not apparently lead to a meaningful effect size in practice. Both CES-D questionnaires exhibit equal validity, reliability, and factor structures and exhibit a little difference in latent means. Therefore, the Internet-based questionnaire represents a promising alternative to the paper-based questionnaire.

Our study addresses the following research questions: Are there differences between handwriting movements on paper and on a tablet computer? Can experienced writers, such as most adults, adapt their graphomotor execution during writing to a rather unfamiliar surface for instance a tablet computer? We examined the handwriting performance of adults in three tasks with different complexity: (a) graphomotor abilities, (b) visuomotor abilities and (c) handwriting. Each participant performed each task twice, once on paper and once on a tablet computer with a pen. We tested 25 participants by measuring their writing duration, in air time, number of pen lifts, writing velocity and number of inversions in velocity. The data were analyzed using linear mixed-effects modeling with repeated measures. Our results reveal differences between writing on paper and on a tablet computer which were partly task-dependent. Our findings also show that participants were able to adapt their graphomotor execution to the smoother surface of the tablet computer during the tasks. (C) 2016 Elsevier B.V. All rights reserved.

Summary Objectives: To evaluate the efficiency and acceptance of digital pen and paper technology for real-time clinical data acquisition. Methods: A prospective interventional unblinded study involving consecutive patients in two clinical settings during a defined time-frame was proposed. The first trial was designed as a stress test to evaluate acceptance in a workload-intensive environment. Acceptance was assessed using observations and a satisfaction questionnaire. The second trial was intended to determine the reliability of data acquisition in a controlled environment. Reliability was assessed by comparing the performance of the digital pen against scanner analysis and a double human blinded acquisition. Results: Overall, users were satisfied with the use of the digital pen (median 3 on a Likert-scale (–5, 5)). Without any specific user training, successful data acquisition was greater than 80%. Use of this technology required less adaptation than standard computer devices, and was easy to learn and use. Ergonomic problems shaded the perception of the technology by inducing an increased cognitive load. Digitalized data was missing either because of a bug or due to lack of data validation. The reliability obtained with the digital pen was significantly lower to that obtained with the scanner. Conclusions: Natural technology such as the digital pen proved to be a good tool in stressful clinical environments without interfering with the normal workload or increasing the time for data acquisition. However, in order to improve quality of data acquisition, designing acquisition forms specifically for the use of digital pens is of paramount importance.

Funding: TR Tech, Surfnet, GRAND, AITF, CFI, NSERC and Smart Technologies. With new computer technologies portable devices are rapidly approaching the dimensions and characteristics of traditional pen and paper-based tools. Text and graphic documents are now commonly viewed using small tablet computers. We conducted a study with small groups of participants to better understand how paper-based text and graphics are used by small collaborative groups as compared to how these groups make use of documents presented on a digital tablet with digital styluses. Our results indicate that digital tools, as compared to paper tools, can affect the levels of verbal communication and participant gaze engagement with other group members. Additionally, we observed how participants spatially arranged paper-based and digital tools during collaborative group activities, how often they switched from digital to paper, and how they still prefer paper overall. Postprint

Professional artists seem to develop a preference for certain types of pencils and paper (Akyuz 1995 ). While theaesthetic outcome may be the obvious reason for this preference, haptic feedback may also play a role in selecting media (Prytherch and Jerrard 2001 , 2003 ). This suggestion implies that artists may be sensitive to vibrotactile signals resulting from the interaction between pencil and paper during drawing. Tribological tests showed that H-range leads generate more resistance than B-range pencils when writing on a standard 80 g/m 2 paper. Moreover, a constant stimulus 2IFC procedure showed that Fine Arts students found it easier to discriminate between H-range and HB than B-range and an HB. This suggests that the human haptic system may be sensitive to the vibrotactile signals generated by the differences in resistance resulting from the interaction of leads with the paper's textured surface. Implications for pencil manufacturing and simulations of virtual writing are discussed. Statement of Relevance: The study assesses the tactile sensitivity of Fine Artists to the physical properties of the interaction between different types of lead and a standard printing paper. The results may help in designing more responsive drawing instruments as well as more realistic virtual drawing interfaces.

Designing computer systems for educational purpose is a difficult task. While many of them have been developed in the past, their use in classrooms is still scarce. We make the hypothesis that this is because those systems take into account the needs of individuals and groups, but ignore the requirements inherent in their use in a classroom. In this work, we present a computer system based on a paper and tangible interface that can be used at all three levels of interaction: individual, group, and classroom. We describe the current state of the interface design and why it is appropriate for classroom orchestration, both theoretically and through two examples for teaching geometry.

The studies reported here examined differences in user behaviour when presented with a low-fidelity paper prototype compared to fully operational product, and evaluated the effectiveness of different types of enhanced labelling of controls. In the first study with a paper prototype, 30 users of high-pressure washers were asked to choose the settings of the temperature control for different cleaning objects, comparing standard with information-enriched control labelling. In the second study, 34 users operated a real high-pressure washer with different forms of control labelling. The results of both studies provided evidence for some benefits of an information-enriched control labelling over traditional temperature-centred controls labelling. Furthermore, a comparative analysis of the data of the two studies suggested that low-fidelity paper prototypes may have to be used with caution. Therefore, designers need to be aware that the behavioural effects induced by different design modifications may be overestimated when using paper prototypes. The implications of the findings are discussed within the framework of an enlarged concept of fidelity.

This paper describes a suite of routines using IgorPro, a powerful analysis and graphing software package for the Macintosh computer, to enhance the ability to analyze, manipulate, and display data recorded with the Discovery™ acquisition software marketed by DataWave Technologies. The routines are able to time-align fast and slow data channels, and are especially useful for analyses that involve both neural and behavioral data. The software was designed for eyeblink conditioning and vocalization experiments, but it can easily be used for analyzing other types of neurobehavioral data. The data are first prepared on the PC with routines that inspect the header of the data file and translate the data file into a compact binary format that can be read by IgorPro. An option is also available to splice out data from unnecessary portions of an intertrial interval. The new file is then put on the Macintosh computer for display and analysis by IgorPro. These routines enable both neural and behavioral data to be quickly and easily reduced, manipulated, and statistically and graphically summarized.