A5100635681- Software Reliability and Analysis Research
- Adversarial Robustness in Machine Learning
- Safety Systems Engineering in Autonomy
- Risk and Safety Analysis
- Software Testing and Debugging Techniques
- Formal Methods in Verification
- Explainable Artificial Intelligence (XAI)
- Machine Learning and Data Classification
- Machine Learning and Algorithms
- Fault Detection and Control Systems
- Radiomics and Machine Learning in Medical Imaging
- Artificial Intelligence in Healthcare and Education
- MXene and MAX Phase Materials
- Advanced ceramic materials synthesis
- Machine Learning in Healthcare
- Anomaly Detection Techniques and Applications
- Advanced Malware Detection Techniques
- Electronic Packaging and Soldering Technologies
- Bayesian Modeling and Causal Inference
- Autonomous Vehicle Technology and Safety
- Energy Load and Power Forecasting
- Network Security and Intrusion Detection
- Sleep and Work-Related Fatigue
- Meteorological Phenomena and Simulations
- Plant Water Relations and Carbon Dynamics
Tongji University
2024-2025
Tongji Hospital
2024-2025
Shenyang Pharmaceutical University
2025
Chinese Academy of Medical Sciences & Peking Union Medical College
2025
Anhui Medical University
2023-2025
Peking Union Medical College Hospital
2025
Shandong University
2024
University of Liverpool
2020-2024
University of Warwick
2023-2024
Tsinghua University
2013-2024
Abstract Large language models (LLMs) have exploded a new heatwave of AI for their ability to engage end-users in human-level conversations with detailed and articulate answers across many knowledge domains. In response fast adoption industrial applications, this survey concerns safety trustworthiness. First, we review known vulnerabilities limitations the LLMs, categorising them into inherent issues, attacks, unintended bugs. Then, consider if how Verification Validation (V&V)...
This paper focuses on the analysis of application effectiveness integration deep learning and computer vision technologies. Deep achieves a historic breakthrough by constructing hierarchical neural networks, enabling end-to-end feature semantic understanding images. The successful experiences in field provide strong support for training algorithms. tight these two fields has given rise to new generation advanced systems, significantly surpassing traditional methods tasks such as machine...

 In this study, we address the challenging task of biomedical text document classification Cancer Doc Classification, specifically focusing on lengthy research papers related to cancer. Unlike previous that often deals with shorter abstracts and concise summaries, curated a unique dataset comprising documents more extensive content, each exceeding 6 pages in length. To tackle challenge, employed Random Forest Tree method. is powerful ensemble learning technique combines multiple...
Machine Learning, as one of the key technologies in field artificial intelligence, has made significant advancements recent years. This study provides a relatively systematic introduction to machine learning. Firstly, it gives an overview historical development learning, and then focuses on analysis classical algorithms Subsequently, elucidates latest research aiming comprehensively explore applications learning various domains discuss potential future directions.
Deep Reinforcement Learning (DRL) has achieved impressive performance in robotics and autonomous systems (RAS). A key challenge to its deployment real-life operations is the presence of spuriously unsafe DRL policies. Unexplored states may lead agent make wrong decisions that could result hazards, especially applications where DRL-trained end-to-end controllers govern behaviour RAS. This letter proposes a novel quantitative reliability assessment framework for DRL-controlled RAS, leveraging...
Symmetry is crucial for gaining insights into the fundamental properties of physical systems, bringing possibilities in studying exotic phenomena such as quantum phase transitions and ground state entanglement. Here, we experimentally simulate a highly controllable extended Rabi model, capable tuning ultra-strong or deep coupling regime, spin-motion-coupled trapped ion. We observe that phonon driven by model with parity symmetry preserved (broken) would experience double (single) excitation...
Abstract Metal aluminum is usually used to deoxidize as a deoxidant in RH reactor for producing ultra-low carbon Al-killed steel. Studying the process of deoxidization significance forecasting distribution and optimization duration. An inhomogeneous two-phase model simulate circulating process. A multicomponent equation diffuse elements, oxygen. The control source term component reaction Then numerical steel established. following results are concluded: simulation accorded with actual...
There is an urgent societal need to assess whether autonomous vehicles (AVs) are safe enough. From published quantitative safety and reliability assessments of AVs, we know that, given the goal predicting very low rates accidents, road testing alone requires infeasible numbers miles be driven. However, previous analyses do not consider any knowledge prior - which could bring substantial advantages if AV design allows strong expectations before testing. We present a new variant Conservative...
Unmanned aerial vehicles (UAVs) as base stations have attracted great attention in emergency communication networks due to flexible deployment. With the popularization of smart devices, demand for multimedia services is increasing disaster relief. Therefore, it very important significantly improve throughput unmanned vehicle station (UAV-BS) while ensuring quality-of-service (QoS) traffic. In this paper, we consider a UAV-BS serve group users downlink who different statistical delay-bound...
Interpretability of Deep Learning (DL) is a barrier to trustworthy AI. Despite great efforts made by the Explainable AI (XAI) community, explanations lack robustness— indistinguishable input perturbations may lead different XAI results. Thus, it vital assess how robust DL interpretability is, given an method. In this paper, we identify several challenges that state-of-the-art unable cope with collectively: i) existing metrics are not comprehensive; ii) techniques highly heterogeneous; iii)...
As Large Language Models (LLMs) become more integrated into our daily lives, it is crucial to identify and mitigate their risks, especially when the risks can have profound impacts on human users societies. Guardrails, which filter inputs or outputs of LLMs, emerged as a core safeguarding technology. This position paper takes deep look at current open-source solutions (Llama Guard, Nvidia NeMo, Guardrails AI), discusses challenges road towards building complete solutions. Drawing robust...
Enterprise innovation investment is influenced by the actions of subjects, whereas regulating shareholders' equity pledge behavior facilitates and finance but also carries dangers affects enterprise investment. Methods:This paper builds an unbalanced panel model to empirically analyze impact controlling pledges on corporate its heterogeneous characteristics. It looks at moderating role financing constraints mediating incentives, using data from A-share listed companies in China's...
Abstract Background Understanding the risk factors for hematoma expansion (HE) in different regions of intracerebral hemorrhage (ICH) can help development more accurate HE prediction tools and implementing effective clinical treatment interventions. This study aims to investigate patients with lobar deep ICH. Methods A retrospective analysis was conducted on 558 cases primary supratentorial ICH from Tongji Hospital Affiliated University. Patients were categorized into groups. Differential...
With the acceleration of global digitalization, research on Central Bank Digital Currencies (CBDCs) as innovative payment tools has increasingly become a hot topic among financial scholars and policymakers. Particularly, China’s digital yuan (e-CNY), one world’s first CBDCs to be implemented large scale, not only advanced internationalization RMB but also exerted profound impact system. From law perspective, this study analyzes legal regulatory mechanisms cross-border payments involving...
Context: Demonstrating high reliability and safety for safety-critical systems (SCSs) remains a hard problem. Diverse evidence needs to be combined in rigorous way: particular, results of operational testing with other from design verification. Growing use machine learning SCSs, by precluding most established methods gaining assurance, makes even more important supporting claims. Objective: We Autonomous Vehicles (AVs) as current example revisit the problem demonstrating reliability. AVs are...
Strong tropical cyclone (TC) Ockhi occurred in the southeastern Arabian Sea (AS) 2017. greatly changed oceanic conditions and induced large variation chlorophyll-a (Chl-a). The dynamic mechanisms of long-term phytoplankton bloom after passage TC were investigated this study. Prominent surface ocean responses, e.g., decreasing temperature salinity, identified from Argo data by comparing pre- post-conditions TC. A was observed AS within area (11°N-14°N, 67°E-70°E) lasted for seven days....
Large Language Models (LLMs) have exploded a new heatwave of AI for their ability to engage end-users in human-level conversations with detailed and articulate answers across many knowledge domains. In response fast adoption industrial applications, this survey concerns safety trustworthiness. First, we review known vulnerabilities limitations the LLMs, categorising them into inherent issues, attacks, unintended bugs. Then, consider if how Verification Validation (V&V) techniques, which been...