報告題目：ALightweight Deep Learning Solution to mmWave Human Activity Recognition
報告人：Kennesaw State University Shaoen Wu教授
報告時間：2024年11月19日（周二）下午15:30-16：30
報告地點：三牌樓校區科研樓208會議室
報告摘要：
Millimeter wave (mmWave) based human activityrecognition is vital in many smart IoT applications. In practical IoTscenarios, fast and accurate human activity recognition is criticallyimportant. In this talk, we talk about a lightweight deep learning solution tohuman activity recognition based on discrete Fourier transformation. The modelhas a fairly small number of model parameters while offering high accuracy inactivity recognition. The core of the solution is a discrete Fourier transformmodule inside a neural network, which converts the temporal features of mmWaveradar activity data into frequency features before a simple classifier performsactivity recognition. The evaluation demonstrates that the DFT-based networkcan achieve the same accuracy as other traditional neural network models, butwith a very small computational load.
報告人簡介：
Dr. Shaoen Wu is the Department Chair and a full professor ofInformation Technology at Kennesaw State University. He also serves as aSteering Committee Chair of IEEE MMTC. Dr. Wu worked as the State Farm EndowedChair Professor in the School of Information Technology at Illinois StateUniversity, served on the Advisory Council of Scholarship for the Vice Provostfor Research, the Dean's Faculty Advisory Board and the assistant departmentchair of computer science at Ball State University, also worked as an assistantprofessor in the School of Computing at the University of Southern Mississippi,a Staff Scientist at ADTRAN, and a Member of Technical Staff at Bell Labs,Lucent Technologies. He has been a General or TPC Chair for severalinternational conferences, including the CSM of Globecom 2021. Dr. Wu hasdirected research projects of several million dollars funded by US federalagencies and industry.
