Finished a comprehensive Review Report Related to Differential Privacy and Federated Learing.

From Intuition, math to implementation details to tech application, and real-world examples.

Final Presentation.

Small Project.

Math: Use Graph property and KL divergence to derive the objective optimization function, then use optimization algorithm to minimize it, and build the statistical model.

Code Realization: Use Python and Matlab to realize the algorithm, use py to realize data visualization and graph property analysis, choose realworld dataset to do some experiments.

Programming: Python(Networkx, Seaborn), Matlab

Used ML models to solve multi-classification problems. Including KNN, Random Forest and Adaboost model. Task including dataset down sampling, data preprocessing, model training and cross-validation, model testing and performance comparason.

Final Project Report.

Dataset: UCI Dataset - 1999-2008 Diabetes Readmission

Programming: Python(Pandas, Scikitlearn )

Result: Realized Dyanimic Port Communcation between Health center(1 Entity), Students(5 Entities), and Hospitals(3 Entities), HC made decision by avaliablity of health center and the severity of student illness.

Details: Phase 1，Health Center as server, built connection by TCP, hospital send the data to HC, students severity ranking to HC, then HC use these info to make decision; Phase 2, Health Center as client, built connection by UDP, sent decising results to hospitals and students. Packet size: one Student info as a packet, one HC response as a packet

Programming: C++ ( Socket programming)

XXXXXXX

Dataset: Yelp Review Dataset

Programming: Python(Pyspark)，Scala，Docker

XXXXXXXX

API: XXXXXXXX

Programming: Java, Android Studio,

XXXXXXXX

Programming: HTML/CSS/JS(Front-end), Python(Back-end), Flask(Py Framework) ,

API: Google Map API.

Presentation:

Record => Demo Video

Devepost => Team

Result: The Prototype Mechine can climb forward and backward in Powerline, can feel the ice and begin deice, can be controlled by manually(Bluetooth).

Details: The whole prototype can be devided into four blocks. 1. Power Supply block, including circuit design and battery chosen. 2. Microcontroller Block, provide logic and control for the device. 3. Sersor and Bluetooth communication Block. 4. Motor & Driver Block, realized mechanism design and motor related calculations.

Software Use: Mechanical(CAD, Solidworks), Electronic(KiCAD, LTSpice), Control(Arduino Studio, + Bluetooth device/Mobile Device , + Temperature/Distance Sensor )

Prototype Video : Complete test, Component test, Climb in Powerline,

Software Use: MATLAB