A5037295904- Masonry and Concrete Structural Analysis
- Topic Modeling
- Sentiment Analysis and Opinion Mining
- Software Engineering Research
- Text and Document Classification Technologies
- Hygrothermal properties of building materials
- Building materials and conservation
- Dam Engineering and Safety
- Reinforcement Learning in Robotics
- Structural Analysis and Optimization
- Agricultural Engineering and Mechanization
- Natural Fiber Reinforced Composites
- Structural Health Monitoring Techniques
- Water Systems and Optimization
- Metallurgy and Material Forming
- Multimodal Machine Learning Applications
- Geomechanics and Mining Engineering
- Structural Analysis of Composite Materials
- Smart Agriculture and AI
- Biomedical Ethics and Regulation
- Metal Forming Simulation Techniques
- Powder Metallurgy Techniques and Materials
- Grouting, Rheology, and Soil Mechanics
- Hydraulic flow and structures
- AI in Service Interactions
Dalian Jiaotong University
2016-2023
University of Science and Technology of China
2020
Xuzhou University of Technology
2015-2017
Dalian University of Technology
2000-2012
Beijing Agricultural Machinery Research Institute
2012
Recently, sentiment analysis has seen remarkable advance with the help of pre-training approaches. However, knowledge, such as words and aspect-sentiment pairs, is ignored in process pre-training, despite fact that they are widely used traditional In this paper, we introduce Sentiment Knowledge Enhanced Pre-training (SKEP) order to learn a unified representation for multiple tasks. With automatically-mined SKEP conducts masking constructs three knowledge prediction objectives, so embed...
Reward shaping is an effective technique for incorporating domain knowledge into reinforcement learning (RL). Existing approaches such as potential-based reward normally make full use of a given function. However, since the transformation human numeric values often imperfect due to reasons cognitive bias, completely utilizing function may fail improve performance RL algorithms. In this paper, we consider problem adaptively We formulate utilization rewards bi-level optimization problem, where...
Recently, sentiment analysis has seen remarkable advance with the help of pre-training approaches. However, knowledge, such as words and aspect-sentiment pairs, is ignored in process pre-training, despite fact that they are widely used traditional In this paper, we introduce Sentiment Knowledge Enhanced Pre-training (SKEP) order to learn a unified representation for multiple tasks. With automatically-mined SKEP conducts masking constructs three knowledge prediction objectives, so embed...
Understanding user instructions in natural language is an active research topic AI and robotics. Typically, are high-level can be reduced into low-level tasks expressed common verbs (e.g., `take', `get', `put'). For robots understanding such instructions, one of the key challenges to process achieve specified with robots' primitive actions. To address this, we propose novel algorithms by utilizing semantic roles defined dictionaries integrating multiple open knowledge generate task plans....
Offline reinforcement learning (RL) provides a promising approach to avoid costly online interaction with the real environment. However, performance of offline RL highly depends on quality datasets, which may cause extrapolation error in process. In many robotic applications, an inaccurate simulator is often available. data directly collected from cannot be used due well-known exploration-exploitation dilemma and dynamic gap between simulation To address these issues, we propose novel...
The bond behavior between the textile and earth-based matrix determined reinforcement effectiveness of composite systems. This paper presented a pull-out experimental study on glass mesh reinforced matrix. was studied using different development length, spacing size thickness, with total 32 specimens. test results showed that peak force had increased by 31.7% 40.5% 200 mm 300 versus 100 respectively. 16 compared to 10 thickness specimens high strength improvement (9.73%) because elevated...
Aiming at the characteristics of forming difficulty for Semimonocoque, a finite element analysis on precision forging was conducted by using rigid-plastic method based DEFORM-3D software. The metal flow law during process studied. equivalent stress, strain distribution and load-stroke curve were analyzed. numerical simulation results used to optimize die structure technological parameters. research verified valid experiment.
An adjustable foaming machine that consisted of air feeding system, liquid device, and control system was designed to satisfy various demands on foam density diameter in the preparation concrete. The air-liquid ratio adjusted by controlling flow, adjustment achieved. single-foam tube structure adopted realize controls pore size adjusting packing medium tube. By using single chip as digital processing core, communication between frequency converter established through RS485 interface, thereby...
This paper describes MDS: a Multimodal-based Dialog System that supports communication between the hearing impaired and hearing-abled. The system converts sign language to speech, combines speech with gesture lip motion using human face. features of face are derived by doing 3D feature extraction speaker's face, so "virtual face" similar actual speaker. main technologies associated include recognition, synthesis synchrony movement speech. Integration synthesis, virtual provides new way...
We propose a system, named ATVHunter, which can pinpoint the precise vulnerable in-app TPL versions and provide detailed information about vulnerabilities TPLs. two-phase detection approach to identify specific versions. Specifically, we extract Control Flow Graphs as coarse-grained feature match potential TPLs in pre-defined database, then opcode each basic block of CFG fine-grained exact build comprehensive database (189,545 unique with 3,006,676 versions) reference database. Meanwhile,...
Abstract By the use of extended finite element method analysis random response a linear structure to continuous excitation field, in time and space, is presented this paper. The includes formulation for obtaining equivalent node force power spectrum. corresponding computer program has been produced. A sluice gate support shows satisfactory agreement with experiment results.