Comparison between parameter-efficient techniques and full fine-tuning: A case study on multilingual news article classification.

Adapters and Low-Rank Adaptation (LoRA) are parameter-efficient fine-tuning techniques designed to make the training of language models more efficient. Previous results demonstrated that these methods can even improve performance on some classification tasks. This paper complements existing research by investigating how these techniques influence classification performance and computation costs compared to full fine-tuning. We focus specifically on multilingual text classification tasks (genre, framing, and persuasion techniques detection; with different input lengths, number of predicted classes and classification difficulty), some of which have limited training data. In addition, we conduct in-depth analyses of their efficacy across different training scenarios (training on the original multilingual data; on the translations into English; and on a subset of English-only data) and different languages. Our findings provide valuable insights into the applicability of parameter-efficient fine-tuning techniques, particularly for multilabel classification and non-parallel multilingual tasks which are aimed at analysing input texts of varying length.

High-throughput fabrication of soft magneto-origami machines.

Soft magneto-active machines capable of magnetically controllable shape-morphing and locomotion have diverse promising applications such as untethered biomedical robots. However, existing soft magneto-active machines often have simple structures with limited functionalities and do not grant high-throughput production due to the convoluted fabrication technology. Here, we propose a facile fabrication strategy that transforms 2D magnetic sheets into 3D soft magneto-active machines with customized geometries by incorporating origami folding. Based on automated roll-to-roll processing, this approach allows for the high-throughput fabrication of soft magneto-origami machines with a variety of characteristics, including large-magnitude deploying, sequential folding into predesigned shapes, and multivariant actuation modes (e.g., contraction, bending, rotation, and rolling locomotion). We leverage these abilities to demonstrate a few potential applications: an electronic robot capable of on-demand deploying and wireless charging, a mechanical 8-3 encoder, a quadruped robot for cargo-release tasks, and a magneto-origami arts/craft. Our work contributes for the high-throughput fabrication of soft magneto-active machines with multi-functionalities.

Unique Catalytic Mechanism for Ru-Loaded Ternary Intermetallic Electrides for Ammonia Synthesis.

Intermetallic electrides have recently shown their priority as catalyst components in ammonia synthesis and CO2 activation. However, their function mechanism has been elusive since its inception, which hinders the further development of such catalysts. In this work, ternary intermetallic electrides La-TM-Si (TM = Co, Fe, and Mn) were synthesized as hosts of ruthenium (Ru) particles for ammonia synthesis catalysis. Although they have the same crystal structure and possess low work functions commonly, the promotion effects on Ru particles rather differ from each other. The catalytic activity follows the sequence of Ru/LaCoSi > Ru/LaFeSi > Ru/LaMnSi. Furthermore, Ru/LaCoSi exhibits much better catalytic durability than the other two. A combination of experiments and first-principles calculations shows that apparent N2 activation energy on each catalyst is much lower than that over conventional Ru-based catalysts, which suggests that N2 dissociation can be conspicuously promoted by the concerted actions of the specific electronic structure and atomic configuration of intermetallic electride-supported catalysts. The NHx formations proceeded on La are energetically favored, which makes it possible to bypass the scaling relations based on only Ru as the active site. The rate-determining step of Ru/La-TM-Si was identified to be NH2 formation. The transition metal (TM) in La-TM-Si electrides has a significant influence on the metal-support interaction of Ru and La-TM-Si. These findings provide a guide for the development of new and effective catalyst hosts for ammonia synthesis and other hydrogenation reactions.

Text mining occupations from the mental health electronic health record: a natural language processing approach using records from the Clinical Record Interactive Search (CRIS) platform in south London, UK.

OBJECTIVES: We set out to develop, evaluate and implement a novel application using natural language processing to text mine occupations from the free-text of psychiatric clinical notes. DESIGN: Development and validation of a natural language processing application using General Architecture for Text Engineering software to extract occupations from de-identified clinical records. SETTING AND PARTICIPANTS: Electronic health records from a large secondary mental healthcare provider in south London, accessed through the Clinical Record Interactive Search platform. The text mining application was run over the free-text fields in the electronic health records of 341 720 patients (all aged ≥16 years). OUTCOMES: Precision and recall estimates of the application performance; occupation retrieval using the application compared with structured fields; most common patient occupations; and analysis of key sociodemographic and clinical indicators for occupation recording. RESULTS: Using the structured fields alone, only 14% of patients had occupation recorded. By implementing the text mining application in addition to the structured fields, occupations were identified in 57% of patients. The application performed on gold-standard human-annotated clinical text at a precision level of 0.79 and recall level of 0.77. The most common patient occupations recorded were 'student' and 'unemployed'. Patients with more service contact were more likely to have an occupation recorded, as were patients of a male gender, older age and those living in areas of lower deprivation. CONCLUSION: This is the first time a natural language processing application has been used to successfully derive patient-level occupations from the free-text of electronic mental health records, performing with good levels of precision and recall, and applied at scale. This may be used to inform clinical studies relating to the broader social determinants of health using electronic health records.

Classification aware neural topic model for COVID-19 disinformation categorisation.

The explosion of disinformation accompanying the COVID-19 pandemic has overloaded fact-checkers and media worldwide, and brought a new major challenge to government responses worldwide. Not only is disinformation creating confusion about medical science amongst citizens, but it is also amplifying distrust in policy makers and governments. To help tackle this, we developed computational methods to categorise COVID-19 disinformation. The COVID-19 disinformation categories could be used for a) focusing fact-checking efforts on the most damaging kinds of COVID-19 disinformation; b) guiding policy makers who are trying to deliver effective public health messages and counter effectively COVID-19 disinformation. This paper presents: 1) a corpus containing what is currently the largest available set of manually annotated COVID-19 disinformation categories; 2) a classification-aware neural topic model (CANTM) designed for COVID-19 disinformation category classification and topic discovery; 3) an extensive analysis of COVID-19 disinformation categories with respect to time, volume, false type, media type and origin source.

Engineering of a genome-reduced strain Bacillus amyloliquefaciens for enhancing surfactin production.

BACKGROUND: Genome reduction and metabolic engineering have emerged as intensive research hotspots for constructing the promising functional chassis and various microbial cell factories. Surfactin, a lipopeptide-type biosurfactant with broad spectrum antibiotic activity, has wide application prospects in anticancer therapy, biocontrol and bioremediation. Bacillus amyloliquefaciens LL3, previously isolated by our lab, contains an intact srfA operon in the genome for surfactin biosynthesis. RESULTS: In this study, a genome-reduced strain GR167 lacking ~ 4.18% of the B. amyloliquefaciens LL3 genome was constructed by deleting some unnecessary genomic regions. Compared with the strain NK-1 (LL3 derivative, ΔuppΔpMC1), GR167 exhibited faster growth rate, higher transformation efficiency, increased intracellular reducing power level and higher heterologous protein expression capacity. Furthermore, the chassis strain GR167 was engineered for enhanced surfactin production. Firstly, the iturin and fengycin biosynthetic gene clusters were deleted from GR167 to generate GR167ID. Subsequently, two promoters PRsuc and PRtpxi from LL3 were obtained by RNA-seq and promoter strength characterization, and then they were individually substituted for the native srfA promoter in GR167ID to generate GR167IDS and GR167IDT. The best mutant GR167IDS showed a 678-fold improvement in the transcriptional level of the srfA operon relative to GR167ID, and it produced 311.35 mg/L surfactin, with a 10.4-fold increase relative to GR167. CONCLUSIONS: The genome-reduced strain GR167 was advantageous over the parental strain in several industrially relevant physiological traits assessed and it was highlighted as a chassis strain for further genetic modification. In future studies, further reduction of the LL3 genome can be expected to create high-performance chassis for synthetic biology applications.

CogStack - experiences of deploying integrated information retrieval and extraction services in a large National Health Service Foundation Trust hospital.

BACKGROUND: Traditional health information systems are generally devised to support clinical data collection at the point of care. However, as the significance of the modern information economy expands in scope and permeates the healthcare domain, there is an increasing urgency for healthcare organisations to offer information systems that address the expectations of clinicians, researchers and the business intelligence community alike. Amongst other emergent requirements, the principal unmet need might be defined as the 3R principle (right data, right place, right time) to address deficiencies in organisational data flow while retaining the strict information governance policies that apply within the UK National Health Service (NHS). Here, we describe our work on creating and deploying a low cost structured and unstructured information retrieval and extraction architecture within King's College Hospital, the management of governance concerns and the associated use cases and cost saving opportunities that such components present. RESULTS: To date, our CogStack architecture has processed over 300 million lines of clinical data, making it available for internal service improvement projects at King's College London. On generated data designed to simulate real world clinical text, our de-identification algorithm achieved up to 94% precision and up to 96% recall. CONCLUSION: We describe a toolkit which we feel is of huge value to the UK (and beyond) healthcare community. It is the only open source, easily deployable solution designed for the UK healthcare environment, in a landscape populated by expensive proprietary systems. Solutions such as these provide a crucial foundation for the genomic revolution in medicine.