A5032061647- Reinforcement Learning in Robotics
- Advanced Sensor and Energy Harvesting Materials
- Metaheuristic Optimization Algorithms Research
- Domain Adaptation and Few-Shot Learning
- Evolutionary Algorithms and Applications
- Advanced Malware Detection Techniques
- Advanced Clustering Algorithms Research
- Adversarial Robustness in Machine Learning
- Complex Network Analysis Techniques
- Advanced Multi-Objective Optimization Algorithms
- Internet Traffic Analysis and Secure E-voting
- Advanced Materials and Mechanics
- Recommender Systems and Techniques
- Fault Detection and Control Systems
- Natural Language Processing Techniques
- Topic Modeling
- Network Security and Intrusion Detection
- Semiconductor materials and devices
- Advancements in Semiconductor Devices and Circuit Design
- Modular Robots and Swarm Intelligence
- Data Management and Algorithms
- Machine Learning and Data Classification
- Advanced Control Systems Optimization
- Anomaly Detection Techniques and Applications
- Face and Expression Recognition
The University of Queensland
2003-2025
Beihang University
2023-2025
Tsinghua University
2015-2024
Shenyang Aerospace University
2024
Chang'an University
2022-2024
Queensland University of Technology
2006-2024
Zhejiang University
2023-2024
Energy Research Institute
2024
Southern University of Science and Technology
2020-2024
Chongqing University of Posts and Telecommunications
2023-2024
Flexible materials with the ability to be bent, strained, or twisted play a critical role in soft robots and stretchable electronics. Although tremendous efforts are focused on developing new excellent stability, inevitable mechanical damage due long term deformation is still an urgent problem tackled. Here, magnetic healing method based Fe-doped liquid metal (Fe-GaIn) conductive ink via noncontact way proposed. Further, multifunctional flexible electronics designed combined performances of...
Given a set of sparsely distributed sensors in the Euclidean plane, mobile robot is required to visit all download data and finally return its base. The effective range each sensor specified by disk, must at least reach boundary start communication. primary goal optimization this scenario minimize traveling distance robot. This problem can be regarded as special case salesman with neighborhoods (TSPN), which known NP-hard. In paper, we present novel TSPN algorithm for class TSPN, yield...
Trajectory sharing and searching have received significant attentions in recent years. In this paper, we propose investigate a novel problem called User Oriented Search (UOTS) for trip recommendation. contrast to conventional trajectory search by locations (spatial domain only), consider both spatial textual domains the new UOTS query. Given data set, query input contains set of intended places given traveler attributes describing traveler's preference. If is connecting/close specified...
Liquid metal transforms into a porous structure and floats on the water surface with heating.
Abstract With inherent flexibility, high electroconductivity, excellent thermal conductivity, easy printability and biosafety, gallium‐based functional liquid metals (LMs) have been increasingly evaluated for biomedical applications, especially as electronic skin (e‐skin). Extending these versatile materials to more challenging applications is a worthwhile pursuit. To realize precise spatiotemporal multisite tumor treatment under an alternating magnetic field (AMF), oxidized GaIn (O‐GaIn)...
Shape tunable liquid metal nanoparticles were fabricated with characterization of biocompatibility, favorable photothermal conversion efficiency and tumor targeting capability for therapy.
Multimodal hashing, which conducts effective and efficient nearest neighbor search across heterogeneous data on large-scale multimedia databases, has been attracting increasing interest, given the explosive growth of content Internet. Recent multimodal hashing research mainly aims at learning compact binary codes to preserve semantic information by labels. The overwhelming majority these methods are similarity preserving approaches approximate pairwise matrix with Hamming distances between...
Abstract Mechanical forces are crucial for normal living organisms as well formation of tumor microenvironments. However, to date, there rather limited trials regulate the mechanical factors toward treatment or imaging. Here, a synergistic antitumor therapy cryoablation and gallium microparticles (GMs) mediated bomb‐explosion‐like destruction is proposed first time. Moreover, GMs demonstrated enhance T2 magnetic resonance imaging (MRI) effect mediate dual‐mode computerized tomography (CT)...
In unsupervised domain adaptation (UDA), a classifier for the target is trained with massive true-label data from source and unlabeled domain. However, collecting in can be expensive sometimes impractical. Compared to true label (TL), complementary (CL) specifies class that pattern does not belong to, hence, CLs would less laborious than TLs. this article, we propose novel setting where composed of complementary-label data, theoretical bound provided. We consider two cases setting: one only...
MOBA games, e.g., Honor of Kings, League Legends, and Dota 2, pose grand challenges to AI systems such as multi-agent, enormous state-action space, complex action control, etc. Developing for playing games has raised much attention accordingly. However, existing work falls short in handling the raw game complexity caused by explosion agent combinations, i.e., lineups, when expanding hero pool case that OpenAI's limits play a only 17 heroes. As result, full without restrictions are far from...
In the unsupervised open set domain adaptation (UOSDA), target contains unknown classes that are not observed in source domain. Researchers this area aim to train a classifier accurately: 1) recognize data (data with classes) and 2) classify other data. To achieve aim, previous study has proven an upper bound of target-domain risk, difference, as important term bound, is used measure risk on By minimizing shallow can be trained aim. However, if very flexible [e.g., deep neural networks...
Multi-stimulus responsive soft materials with integrated functionalities are elementary blocks for building intelligent systems, but their rational design remains challenging. Here, we demonstrate an architecture sensitized by magnetized liquid metal droplets that dispersed in a highly stretchable elastomer network. The supercooled serve as microscopic latent heat reservoirs, and controllable solidification releases localized thermal energy/information flows enabling programmable...
Abstract Bacterial infections, especially caused by multidrug‐resistant bacteria, pose a big challenge to the healthcare system. As group of historic agents, metals with broad‐spectrum antibacterial activity are regarded as promising alternatives tackle antibiotic resistance. Among them, gallium ions have presented encouraging effects in research and preclinic studies. However, utilization has difficulty achieving high targeting long‐term effectiveness. With renaissance liquid metal, here,...
Abstract Multi-camera depth estimation has gained significant attention in autonomous driving due to its importance perceiving complex environments. However, extending monocular self-supervised methods multi-camera setups introduces unique challenges that existing techniques often fail address. In this paper, we propose STViT+ , a novel Transformer-based framework for estimation. Our key contributions include: 1) the Spatial-Temporal Transformer (STTrans) which integrates local spatial...
Abstract Liquid metals are of great importance in developing wearable devices and soft robotics owing to its high conductivity flexibility. However, the density such turned out be big concern for many practical situations. With generalized purpose, a new conceptual material as lightweight liquid metal entity, which can light water, is proposed here. For illustration, an unconventionally ultralight composed eutectic galliumindium alloys (eGaIn) glass bubbles demonstrated, whose reduced below...
Factorization Machines (FMs) refer to a class of general predictors working with real valued feature vectors, which are well-known for their ability estimate model parameters under significant sparsity and have found successful applications in many areas such as the click-through rate (CTR) prediction. However, standard FMs only produce single fixed representation each across different input instances, may limit CTR model’s expressive predictive power. Inspired by success Input-aware (IFMs),...
ConspectusConventional robots can accomplish defined tasks but often encounter troubles when handling irregular objects under unstructured environments. Soft robots, with supercompliance, large transformation, and high environmental adaptability, hold big promise for delicate manipulations such as grasping soft or delivering precious biomedical samples. Even a step further, if are endowed the extraordinary behaviors to freely transform among different morphologies constructions just like...