Self-Attention Based Vision Processing for Prosthetic Vision.

We investigate Self-Attention (SA) networks for directly learning visual representations for prosthetic vision. Specifically, we explore how the SA mechanism can be leveraged to produce task-specific scene representations for prosthetic vision, overcoming the need for explicit hand-selection of learnt features and post-processing. Further, we demonstrate how the mapping of importance to image regions can serve as an explainability tool to analyse the learnt vision processing behaviour, providing enhanced validation and interpretation capability than current learning-based methods for prosthetic vision. We investigate our approach in the context of an orientation and mobility (OM) task, and demonstrate its feasibility for learning vision processing pipelines for prosthetic vision.

A fluorescent photoaffinity probe for formyl peptide receptor 1 labelling in living cells.

Fluorescent ligands for G-protein coupled receptors (GPCRs) are valuable tools for studying the expression, pharmacology and modulation of these therapeutically important proteins in living cells. Here we report a fluorescent photoaffinity probe for Formyl peptide receptor 1 (FPR1), a critical component of the innate immune response to bacterial infection and a promising target in inflammatory diseases. We demonstrate that the probe binds and covalently crosslinks to FPR1 with good specificity at nanomolar concentrations in living cells and is a useful tool for visualisation and characterisation of this receptor.

Pseudomonas aeruginosa utilizes the host-derived polyamine spermidine to facilitate antimicrobial tolerance.

Pseudomonas aeruginosa undergoes diversification during infection of the cystic fibrosis (CF) lung. Understanding these changes requires model systems that capture the complexity of the CF lung environment. We previously identified loss-of-function mutations in the 2-component regulatory system sensor kinase gene pmrB in P. aeruginosa from CF lung infections and from experimental infection of mice. Here, we demonstrate that, while such mutations lowered in vitro minimum inhibitory concentrations for multiple antimicrobial classes, this was not reflected in increased antibiotic susceptibility in vivo. Loss of PmrB impaired aminoarabinose modification of LPS, increasing the negative charge of the outer membrane and promoting uptake of cationic antimicrobials. However, in vivo, this could be offset by increased membrane binding of other positively charged molecules present in lungs. The polyamine spermidine readily coated the surface of PmrB-deficient P. aeruginosa, reducing susceptibility to antibiotics that rely on charge differences to bind the outer membrane and increasing biofilm formation. Spermidine was elevated in lungs during P. aeruginosa infection in mice and during episodes of antimicrobial treatment in people with CF. These findings highlight the need to study antimicrobial resistance under clinically relevant environmental conditions. Microbial mutations carrying fitness costs in vitro may be advantageous during infection, where host resources can be utilized.

Virtual Traffic Light Implementation on a Roadside Unit over 802.11p Wireless Access in Vehicular Environments.

Blind intersections have high accident rates due to the poor visibility of oncoming traffic, high traffic speeds, and lack of infrastructure (e.g., stoplights). These intersections are more commonplace in rural areas, where traffic infrastructure is less developed. The Internet of Vehicles (IoV) aims to address such safety concerns through a network of connected and autonomous vehicles (CAVs) that intercommunicate. This paper proposes a Road-Side Unit-based Virtual Intersection Management (RSU-VIM) over 802.11p system consisting of a Field-Programmable Gate Array (FPGA) lightweight RSU that is solar power-based and tailored to rural areas. The RSU utilizes the proposed RSU-VIM algorithm adapted from existing virtual traffic light methodologies to communicate with vehicles over IEEE 802.11p and facilitate intersection traffic, minimizing visibility issues. The implementation of the proposed system has a simulated cloud delay of 0.0841 s and an overall system delay of 0.4067 s with 98.611% reliability.

Deep reinforcement learning for task-based feature learning in prosthetic vision.

Developing hand-crafted visual features to enhance perception with prosthetic vision devices can often miss important aspects of a given task. Retinal implants suffer from the need to create low-dimensional features for elaborate tasks such as navigation in complex environments. Using Deep Reinforcement Learning (DRL), visual features are learnt through task-based simulations that remove the ambiguity of inferring the visual information most crucial to a specific activity. Learning task-based features ensures that the visual information is salient to the tasks an implant recipient may be undertaking and eliminates potentially redundant features. In this paper, we focus specifically on basic orientation and mobility, and the methods for feature learning and visualisation in structured 3D environments. We propose a new model for learning visual features through task-based simulations and show that learnt features can be transferred directly to real RGB-D images. We demonstrate this new scalable approach for feature learning in simulation and open the possibility for more complex simulations of more complex tasks in the future.

A trait-interpersonal perspective on suicide risk in criminal offenders.

Despite elevated rates of suicide among offenders, research has yet to adequately address theoretically driven risk models in this population. The present study addresses such a gap by investigating a synthesized framework using 2 well-developed theoretical models, the Five-Factor Model (FFM) of personality and the interpersonal-psychological theory of suicide (IPTS). Archival data from mitigation evaluations for pre-sentenced criminal offenders (n = 307) were analyzed using structural equation modeling (SEM). Results supported a hypothesized personality-IPTS-suicide framework in which neuroticism and extraversion predict IPTS components, namely perceived burdensomeness, thwarted belongingness, and acquired capability. IPTS constructs subsequently directly and indirectly (via suicidal ideation) predicted suicide potential. Findings have important implications for suicide theory, research, and risk assessment with offenders.