Yu-Chuan Hung

Reduced ticket resolution time by 20% and improved network access control by automating processes with PowerShell scripting, decreasing manual workload and boosting response efficiency., Strengthened security infrastructure by implementing VPN setup and Malwarebytes for malware protection, increasing remote access security and reducing vulnerability risk by 25%., Enhanced ERP system efficiency through optimization of data processing and management workflows using ASP and MSSQL, resulting in streamlined operations and significantly reduced processing time.

Improved transaction per second (TPS) rate by 1.5x through algorithmic enhancements, boosting system efficiency., Integrated Hyperledger Fabric to create a hybrid database, focusing on security, scalability, and performance., Conducted performance tests using Caliper to assess throughput, latency, and resource utilization.

Deployed 5G Core Network via open5gs and free5gc with the assistance of UERANSIM to conduct connections to 5G core networks. Configured network slices to simulate 5G core networks for 5G security research., Deployed free5gmano to manage and schedule 5G network slices with docker and Kubernetes.

Developed prototype and end-to-end pipeline to visualize errors between data loss and frequency for improvement with the Fast Fourier Transform (FFT) method, improving model with 80 percent accuracy., Analyzed motors of robotic arms and examined inner and outer bearing faults and frequency of the occurrence., Advised the company on updating the vibration sensor via analyzing anomaly signals to prove the modification.

Developed prototype of Raspberry Pi Cluster using Message Passing Interface (MPI) and implemented OpenCV for real-time traffic monitoring, achieving 1.5-2x faster-processing speed., Teamed and earned an Excellence Award at the competition (Machine Learning, top 10 among 20 groups) by implementing the model to classify human emotions with 8 categories with Raspberry Pi., Evaluated the model performance and improved 10 percent model accuracy with Raspberry Pi Cluster.

Processed EEG motor imagery signals by MATLAB to synthesize alpha and beta bands of brain channels., Implemented Neural Network and Convolutional Neural Network by Keras and SciPy to train the model, achieving a 15 percent improvement in accuracy. Analyzed the association between prediction and ground truth via Matplotlib., Fine-tuned different Short-Time Fourier Transform (STFT) window sizes to optimize data. Adjusted and analyze layers and neurons to optimize models and determine ideal potential activation functions and optimization.