Current rates of habitat degradation and climate change are causing unprecedented declines in global biodiversity. Studies on vertebrates highlight how conservation genomics can be effective in identifying and managing threatened populations, but it is unclear how vertebrate-derived metrics of genomic erosion translate to invertebrates, with their markedly different population sizes and life histories. The Black-veined White butterfly (Aporia crataegi) was extirpated from Britain in the 1920s. Here, we sequenced historical DNA from 17 specimens collected between 1854 and 1924 to reconstruct demography and compare levels of genomic erosion between extirpated British and extant European mainland populations. We contrast these results using modern samples of the Common Blue butterfly (Polyommatus icarus); a species with relatively stable demographic trends in Great Britain. We provide evidence for bottlenecks in both these species around the period of post-glacial colonization of the British Isles. Our results r…
SME innovation plays a central role in achieving regional economic development, with firms led by minority groups making significant contributions to the UK economy. However, access to funding still presents a major barrier for underrepresented founders, impeding their engagement in innovation and resulting in missed opportunities to promote sustainable and inclusive economic growth. Our study advances knowledge on the accessibility of innovation funding by adopting an intersectional lens to examine the impact of founders’ social identities (i.e. gender, ethnicity and education) on their success in obtaining funding. It draws on innovation funding applications data submitted by founders of small and micro businesses between 2016 and 2023. We make the following contributions. Our findings indicate that despite being underrepresented, women founders who apply for innovation funding are as successful as men. Additionally, the heterogeneity of underrepresented founders and specific intersectional challenges influ…
The elimination of micropollutants such as highly fluorinated substances, including per- and polyfluoroalkyl substances (PFAS), in wastewater treatment plants has been receiving growing attention due to the urgent need to minimize their adverse effects on natural water and associated ecosystems. Conventional treatment methods often fall short in effectively removing PFAS. In this study, the Hydrodynamic Cavitation on a Chip concept (HCOC) was utilized to degrade 11 common PFAS variants (PFAS11) for the first time in three different hydrodynamic cavitation reactor set-ups, each enhanced with surface modifications involving roughness elements. Stockholm municipal wastewater treated by a Membrane BioReactor (MBR) process was subjected to fully developed cavitating flow treatment using the three distinct microscale hydrodynamic cavitation (HC) reactors. The obtained results indicate that the chemical-free HCOC technique employed in this study has a significant potential in the degradation of nearly all investigat…
Course materials for the taught days September 2024
Background: Many cancer survivors following primary treatment have prolonged poor quality of life. Aim: To determine the effectiveness of a bespoke digital intervention to support cancer survivors. Design: Pragmatic parallel open randomised trial. Setting: UK general practices. Methods: People having finished primary treatment (<= 10 years previously) for colo-rectal, breast or prostate cancers, with European-Organization-for-Research-and-Treatment-of-Cancer QLQ-C30 score <85, were randomised by online software to: 1)detailed ‘generic’ digital NHS support (‘LiveWell’;n=906), 2) a bespoke complex digital intervention (‘Renewed’;n=903) addressing symptom management, physical activity, diet, weight loss, distress, or 3) ‘Renewed-with-support’ (n=903): ‘Renewed’ with additional brief email and telephone support. Results: Mixed linear regression provided estimates of the differences between each intervention group and generic advice: at 6 months (primary time point: n’s respectively 806;749;705) all groups improve…
With the rapid advance of machine learning (ML) technology, large language models (LLMs) are increasingly explored as an intelligent tool to generate program code from natural language specifications. However, existing evaluations of LLMs have focused on their capabilities in comparison with humans. It is desirable to evaluate their usability when deciding on whether to use a LLM in software production. This paper proposes a user centric method for this purpose. It includes metadata in the test cases of a benchmark to describe their usages, conducts testing in a multi-attempt process that mimics the uses of LLMs, measures LLM generated solutions on a set of quality attributes that reflect usability, and evaluates the performance based on user experiences in the uses of LLMs as a tool. The paper also reports a case study with the method in the evaluation of ChatGPT’s usability as a code generation tool for the R programming language. Our experiments demonstrated that ChatGPT is highly useful for generating R p…
In the scenario-based evaluation of machine learning models, a key problem is how to construct test datasets that represent various scenarios. The methodology proposed in this paper is to construct a benchmark and attach metadata to each test case. Then a test system can be constructed with test morphisms that filter the test cases based on metadata to form a dataset. The paper demonstrates this methodology with large language models for code generation. A benchmark called ScenEval is constructed from problems in textbooks, an online tutorial website and Stack Overflow. Filtering by scenario is demonstrated and the test sets are used to evaluate ChatGPT for Java code generation. Our experiments found that the performance of ChatGPT decreases with the complexity of the coding task. It is weakest for advanced topics like multi-threading, data structure algorithms and recursive methods. The Java code generated by ChatGPT tends to be much shorter than reference solution in terms of number of lines, while it is mo…
With the rapid development of Large Language Models (LLMs), a large number of machine learning models have been developed to assist programming tasks including the generation of program code from natural language input. However, how to evaluate such LLMs for this task is still an open problem despite of the great amount of research efforts that have been made and reported to evaluate and compare them. This paper provides a critical review of the existing work on the testing and evaluation of these tools with a focus on two key aspects: the benchmarks and the metrics used in the evaluations. Based on the review, further research directions are discussed.
Memristors, typically considered for their non-volatile resistive memory for high-density memory designs, have shown very good sensitivity to various chemicals. As such, these devices can also be fabricated as chemical sensors with intrinsic memory. When fabricated for sensing chemicals, the switching state of the devices, depending on the amount of the applied bias voltage/current, also changes in the presence of the chemicals, compared to when the chemicals are not present. We have observed that this property can be combined with the device’s intrinsic memory to directly digitise sensed information. To this end, in this paper, we propose an innovative technique to directly digitise the sensor readings, e.g. gas concentrations, simply by pulsing the devices to exploit the memory behaviour in the presence of a chemical to change the state of the device. Essentially, we are relying on the sensors to tell us when a certain property of a chemical is sensed by switching its state while this information is digitis…