School of Engineering Department of Computer Science and Engineering 90 AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN, Gary Shueng Han / CSE Student: XIA, Zhiqiu / DSCT Course: UROP3100, Fall This semester, I enrolled in UROP 3100 to continue my work on the project AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data, supervised by Prof. Chan. In the first part, I mainly introduced the work on PDR data analysis continuing from the last semester. We leveraged the property of randomness of dropout and tried to use such uncertainty to improve the prediction accuracy. In the next part, I mainly introduced the work about source-free domain adaptation. We leveraged the uncertainty of the model and tried to use the distribution behind it to adapt the pre-trained model for the new data. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN, Gary Shueng Han / CSE Student: YANG, Lin / COMP Course: UROP3100, Fall UROP4100, Spring Computed tomography perfusion can provide information on blood flow in the brain, which can help doctors diagnose strokes and other disorders of brain tissue. The auxiliary software currently used in clinical practice requires doctors to mark the arterial input function point before generating the parameter map required for diagnosis. This process takes a long time, is costly, and requires a lot of labor. Therefore, in this paper, we try to propose a method for predicting feature maps based on CT perfusion maps, thereby reducing the workload of doctors, shortening the diagnosis time, and reducing the cost of medical treatment. In addition, our method also selects keyframes to achieve good predictions with a very small number of original images. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN, Gary Shueng Han / CSE Student: YU, Chun Ho / ELEC Course: UROP1100, Summer The proposed algorithm in Paper 1 uses the combined features of accelerometer, magnetometer, and gyroscope data from an IMU sensor for position estimation. The pitch and roll values are estimated based on a fusion of accelerometer and gyroscope sensor values. The estimated pitch values are used for step detection, and the step lengths are estimated by using the pitching amplitude. The heading of the pedestrian is estimated by the fusion of magnetometer and gyroscope sensor values. Finally, the position is estimated based on the step length and heading information. The experimental results show that the proposed algorithm achieves a high position accuracy that significantly outperforms that of conventional estimation methods used for validation. The algorithm achieved an average accuracy of 1.03 meters in position estimation.
RkJQdWJsaXNoZXIy NDk5Njg=