A5060167899- Recommender Systems and Techniques
- Ethics and Social Impacts of AI
- Advanced Bandit Algorithms Research
- Mobile Crowdsensing and Crowdsourcing
- Non-Invasive Vital Sign Monitoring
- Privacy-Preserving Technologies in Data
- Explainable Artificial Intelligence (XAI)
- Advanced Sensor and Energy Harvesting Materials
- Consumer Market Behavior and Pricing
- Smart Grid Security and Resilience
- Distributed and Parallel Computing Systems
- Privacy, Security, and Data Protection
- Context-Aware Activity Recognition Systems
- Service-Oriented Architecture and Web Services
- Gas Sensing Nanomaterials and Sensors
- Adversarial Robustness in Machine Learning
- Human Mobility and Location-Based Analysis
University of Manchester
2021-2022
Virginia Tech
2019-2021
Harbin Institute of Technology
2017
Weihai Science and Technology Bureau
2017
Cyber-physical systems U+0028 CPS U+0029 are complex with organic integration and in-depth collaboration of computation, communications control 3C technology. Subject to the theory technology existing network physical systems, development is facing enormous challenges. This paper first introduces concept characteristics analyzes present situation researches. Then discussed from perspectives system model, information processing software design. At last it main obstacles key researches in...
We study fairness in collaborative-filtering recommender systems, which are sensitive to discrimination that exists historical data. Biased data can lead methods make unfair predictions for users from minority groups. identify the insufficiency of existing metrics and propose four new address different forms unfairness. These be optimized by adding terms learning objective. Experiments on synthetic real show our better measure than baseline, objectives effectively help reduce
We study fairness in collaborative-filtering recommender systems, which are sensitive to discrimination that exists historical data. Biased data can lead collaborative filtering methods make unfair predictions against minority groups of users. identify the insufficiency existing metrics and propose four new address different forms unfairness. These be optimized by adding terms learning objective. Experiments on synthetic real show our better measure than baseline, objectives effectively help reduce
Imagine a food recommender system -- how would we check if it is \emph{causing} and fostering unhealthy eating habits or merely reflecting users' interests? How much of user's experience over time with caused by the system's choices biases, based on preferences biases? Popularity bias filter bubbles are two most well-studied but prior research has focused understanding behavior in single recommendation step. do these biases interplay user behavior, what types experiences created from...
Despite the importance of respiration and metabolism measurement in daily life, they are not widely available to ordinary people because sophisticated expensive equipment. Here, we first report a straightforward economical approach monitoring respiratory function metabolic rate using wearable piezoelectric airflow transducer (WPAT). A self-shielded bend sensor is designed by sticking two uniaxially drawn poly l-lactic acid films with different cutting angles, then mounted on one end plastic...
Large volumes of data allow for modern application statistical and mathematical models to practical social issues. Many applications predictive like criminal activity heat mapping, recidivism estimation, child safety scoring rely on that may be incomplete, incorrect, or biased. sensitive historical issues can unintentionally incorporated into predictions causing ethical mistreatment. This work proposes a mechanism continuously mitigating model bias by using algorithms produce from reasonably...