FedTKD: A Trustworthy Heterogeneous Federated Learning Based on Adaptive Knowledge Distillation.

Federated learning allows multiple parties to train models while jointly protecting user privacy. However, traditional federated learning requires each client to have the same model structure to fuse the global model. In real-world scenarios, each client may need to develop personalized models based on its environment, making it difficult to perform federated learning in a heterogeneous model environment. Some knowledge distillation methods address the problem of heterogeneous model fusion to some extent. However, these methods assume that each client is trustworthy. Some clients may produce malicious or low-quality knowledge, making it difficult to aggregate trustworthy knowledge in a heterogeneous environment. To address these challenges, we propose a trustworthy heterogeneous federated learning framework (FedTKD) to achieve client identification and trustworthy knowledge fusion. Firstly, we propose a malicious client identification method based on client logit features, which can exclude malicious information in fusing global logit. Then, we propose a selectivity knowledge fusion method to achieve high-quality global logit computation. Additionally, we propose an adaptive knowledge distillation method to improve the accuracy of knowledge transfer from the server side to the client side. Finally, we design different attack and data distribution scenarios to validate our method. The experiment shows that our method outperforms the baseline methods, showing stable performance in all attack scenarios and achieving an accuracy improvement of 2% to 3% in different data distributions.

Natural Language Generation and Understanding of Big Code for AI-Assisted Programming: A Review.

This paper provides a comprehensive review of the literature concerning the utilization of Natural Language Processing (NLP) techniques, with a particular focus on transformer-based large language models (LLMs) trained using Big Code, within the domain of AI-assisted programming tasks. LLMs, augmented with software naturalness, have played a crucial role in facilitating AI-assisted programming applications, including code generation, code completion, code translation, code refinement, code summarization, defect detection, and clone detection. Notable examples of such applications include the GitHub Copilot powered by OpenAI's Codex and DeepMind AlphaCode. This paper presents an overview of the major LLMs and their applications in downstream tasks related to AI-assisted programming. Furthermore, it explores the challenges and opportunities associated with incorporating NLP techniques with software naturalness in these applications, with a discussion on extending AI-assisted programming capabilities to Apple's Xcode for mobile software development. This paper also presents the challenges of and opportunities for incorporating NLP techniques with software naturalness, empowering developers with advanced coding assistance and streamlining the software development process.

A rule-based energy management scheme for long-term optimal capacity planning of grid-independent microgrid optimized by multi-objective grasshopper optimization algorithm.

Off-grid electrification of remote communities using sustainable energy systems (SESs) is a requisite for realizing sustainable development goals. Nonetheless, the capacity planning of the SESs is challenging as it needs to fulfil the fluctuating demand from a long-term perspective, in addition to the intermittency and unpredictable nature of renewable energy sources (RESs). Owing to the nonlinear and non-convex nature of the capacity planning problem, an efficient technique must be employed to achieve a cost-effective system. Existing techniques are, subject to some constraints on the derivability and continuity of the objective function, prone to premature convergence, computationally demanding, follows rigorous procedures to fine-tune the algorithm parameters in different applications, and often do not offer a fair balance during the exploitation and exploration phase of the optimization process. Furthermore, the literature review indicates that researchers often do not implement and examine the energy management scheme (EMS) of a microgrid while computing for the capacity planning problem of microgrids. This paper proposes a rule-based EMS (REMS) optimized by a nature-inspired grasshopper optimization algorithm (GOA) for long-term capacity planning of a grid-independent microgrid incorporating a wind turbine, a photovoltaic, a battery (BT) bank and a diesel generator ( D g e n ). In which, a rule-based algorithm is used to implement an EMS to prioritize the usage of RES and coordinate the power flow of the proposed microgrid components. Subsequently, an attempt is made to explore and confirm the efficiency of the proposed REMS incorporated with GOA. The ultimate goal of the objective function is to minimize the cost of energy (COE) and the deficiency of power supply probability (DPSP). The performance of the REMS is examined via a long-term simulation study to ascertain the REMS resiliency and to ensure the operating limit of the BT storage is not violated. The result of the GOA is compared with particle swarm optimization (PSO) and a cuckoo search algorithm (CSA). The simulation results indicate that the proposed technique's superiority is confirmed in terms of convergence to the optimal solution. The simulation results confirm that the proposed REMS has contributed to better adoption of a cleaner energy production system, as the scheme significantly reduces fuel consumption, CO 2 emission and COE by 92.4%, 92.3% and 79.8%, respectively as compared to the conventional D g e n . The comparative evaluation of the algorithms shows that REMS-GOA yields a better result as it offers the least COE (objective function), at $0.3656/kW h, as compared to the REMS-CSA at $0.3662/kW h and REMS-PSO at $0.3674/kW h, for the desired DPSP of 0%. Finally, sensitivity analysis is performed to highlight the effect of uncertainties on the system inputs that may arise in the future.

A comprehensive review of demand side management in distributed grids based on real estate perspectives.

A major challenge in renewable energy planning and integration with existing systems is the management of intermittence of the resources and customer demand uncertainties that are attributed to climates. In emerging distributed grids, state-of-the-art optimization techniques were used for cost and reliability objectives. In the existing literature, power dispatch and demand side management schemes were implemented for various techno-economic objectives. In renewable energy-based distributed grids, power dispatch is strategic to system operations. However, demand side management is preferred, as it allows more options for customer participation and active management of energy in buildings. Moreover, the demand side management can simply follow supplies. This paper investigates the implications of demand side management as it affects planning and operations in renewable energy-based distributed grids. Integration of demand side management in customer-oriented plans such as the time-of-use and real-time-pricing on residential and commercial demands is conceptualised to ensure effective customer participation which maintains the valued comforts. Moreover, the optimised tariff integrated demand side management implementations based on the utility-initiated demand response programmes are envisaged to offset conflicting objectives of the economy and customer comforts within residential and commercial demands and are also viewed as a step towards efficient management of energy in buildings.

MEGA: Machine Learning-Enhanced Graph Analytics for Infodemic Risk Management.

The COVID-19 pandemic brought not only global devastation but also an unprecedented infodemic of false or misleading information that spread rapidly through online social networks. Network analysis plays a crucial role in the science of fact-checking by modeling and learning the risk of infodemics through statistical processes and computation on mega-sized graphs. This article proposes MEGA, Machine Learning-Enhanced Graph Analytics, a framework that combines feature engineering and graph neural networks to enhance the efficiency of learning performance involving massive graphs. Infodemic risk analysis is a unique application of the MEGA framework, which involves detecting spambots by counting triangle motifs and identifying influential spreaders by computing the distance centrality. The MEGA framework is evaluated using the COVID-19 pandemic Twitter dataset, demonstrating superior computational efficiency and classification accuracy.

Intertrochanteric fracture with distal extension: When is the short proximal femoral nail antirotation too short?

INTRODUCTION: The lesser trochanter (LT) fragment in the multifragmentary intertrochanteric femur fracture (AO 31A2.2) may extend distally. If the fragment extends too distally, fixation with a short proximal femoral nail antirotation (PFNA-II) device may not be sufficient. The exact length of distal extension that can be tolerated by the short PFNA-II is not known, therefore it is our objective to determine it. MATERIALS AND METHODS: A finite element analysis was performed on AO 31A2.2 fracture fixed with a 200mm length size 10 PFNA-II. The construct was loaded vertically to clinical failure of 10mm displacement. This was repeated with the size of the LT fragment increasing distally at intervals, up to 120mm from the base of the LT. The process was also repeated with the bone properties substituted with osteoporotic properties. The stiffness, maximum vertical reaction force, and the plastic deformation area were investigated. RESULTS: In both non-osteoporotic and osteoporotic model, the stiffness and the maximum vertical reaction force of the construct dropped significantly when the LT fragment is larger than 40mm. Beyond 40mm of LT fragment size, there was a rapid increase in the area of plastic deformation of the cortical bone distal to the intertrochanteric fracture, signifying structural failure of the construct. CONCLUSION: A long PFNA-II should be considered when fixing a multifragmentary intertrochanteric fracture if the LT fragment extends 40mm distal to the distal base of the LT as the construct fails rapidly upon uniaxial load to failure. Clinically, this threshold may be smaller to account for the multi-axial and dynamic stresses.

Multi-channel silicon photonic receiver based on ring-resonators.

We demonstrated a high performance monolithically integrated multi-channel receiver fabricated on the SOI platform. This receiver is composed of a 1 x 8 Si-based ring-resonators filter and an array of high speed waveguided Ge-on-Si photodetectors. The optical channel spacing is about 1.5 nm. The responsivity of Ge-on-Si photodetector is about 1.0 A/W at the wavelength range of 1554 nm to 1564 nm. Each channel is capable of operating at a data rate of 20 Gbps, resulting in an aggregate data rate of 160 Gbps. At a BER of 1 x 10(-11), the receiver showed an optical input sensitivity of between -20 dBm and -21 dBm for each channel at 10 Gbps data rate.

Suspended optical fiber-to-waveguide mode size converter for silicon photonics.

In this paper, an efficient and novel optical fiber-to-waveguide mode size converter for Si Photonics devices with sub-micron waveguides is developed on the SOI platform. This optical converter is composed of a suspended SiO(2) waveguide and overlapped Si nano-tapers located in the center of suspended SiO(2) waveguide. Laterally connected SiO(2) beams provide structural support for the suspended SiO(2) waveguide. The optical input signal from the optical fiber is launched into the suspended SiO(2) waveguide, and then coupled into the Si nano-taper. The measured coupling loss using a lensed fiber with 5 microm spot diameter is 1.7 ~2.0 dB/facet for TE mode and 2.0 ~2.4 dB/facet for TM mode in the wavelength range of 1520 ~1600 nm. When a cleaved fiber with 9.2microm spot diameter is used, the coupling losses for both TE and TM modes are less than 4.0 dB/facet at 1550 nm. For the case of lensed fiber, the alignment tolerances for both TE and TM modes are about +/- 1.7 microm for 1 dB excess loss in both X and Y axes. The alignment tolerances for both modes of TE and TM are relaxed, exceeding +/- 2.5 microm for 1 dB excess loss in both X and Y axes when a cleaved fiber is used.

Obstructive sleep apnoea and Type 2 diabetes mellitus: are they connected?

Obstructive sleep apnoea (OSA), a sleep-related breathing condition, is diagnosed based on a patient's apnoea-hypopnea index from a sleep study, and the presence or absence of symptoms. Diabetes mellitus (DM) and OSA share a significant common risk factor, obesity, with all three conditions contributing to the risk of developing cardiovascular diseases. The pathophysiological links between OSA and DM are still unclear, but intermittent hypoxia may be an important mechanism. More awareness of the possible link between OSA and DM is needed, given their increasing prevalence locally and worldwide. Continuous positive airway pressure is the standard treatment for OSA, while weight loss through dietary and lifestyle modifications is important to holistically manage patients with either condition. There is currently insufficient evidence to support the benefits of screening every diabetic patient for OSA. However, diabetic patients with symptoms suggestive of OSA should be referred to a sleep specialist for further evaluation.

Urinary tract infections in adults.

A urinary tract infection (UTI) is a collective term for infections that involve any part of the urinary tract. It is one of the most common infections in local primary care. The incidence of UTIs in adult males aged under 50 years is low, with adult women being 30 times more likely than men to develop a UTI. Appropriate classification of UTI into simple or complicated forms guides its management and the ORENUC classification can be used. Diagnosis of a UTI is based on a focused history, with appropriate investigations depending on individual risk factors. Simple uncomplicated cystitis responds very well to oral antibiotics, but complicated UTIs may require early imaging, and referral to the emergency department or hospitalisation to prevent urosepsis may be warranted. Escherichia coli remains the predominant uropathogen in acute community-acquired uncomplicated UTIs and amoxicillin-clavulanate is useful as a first-line antibiotic. Family physicians are capable of managing most UTIs if guided by appropriate history, investigations and appropriate antibiotics to achieve good outcomes and minimise antibiotic resistance.

Microfabricated endoscopic probe integrated MEMS micromirror for optical coherence tomography bioimaging.

In this paper, we present a miniaturized endoscopic probe, consisted of MEMS micromirror, silicon optical bench (SiOB), grade index (GRIN) lens, single mode optical fiber (SMF) and transparent housing, for optical coherence tomography (OCT) bioimaging. Due to the use of the MEMS micromirror, the endoscopic OCT system is highly suitable for non-invasive imaging diagnosis of a wide variety of inner organs. The probe engineering and proof of concept were demonstrated by obtaining the two-dimensional OCT images with a cover slide and an onion used as standard samples and the axial resolution was around 10µm.