Allen Chien
Multi Agent Reinforcement LearningRobotics
Experience
- Undergraduate Research Assistant @ BadgerRL
- Proposal Writing Intern @ L'SPACE NASA
- REU student @ DAIR, Hunter College(CUNY)
- Supplemental Instruction Tutor @ LaGuardia CC
- Undergraduate Research Assistant @ CUNY LaGCC
- Information Technology Intern @ NeedTech Inc.
- NCAS - National Community College Aerospace Scholar @ NASA
Publications
- "Email Feature Classification and Analysis of Phishing Email Detection Using Machine Learning Techniques" @ IEEE CSDE2023
Honors & Rewards
Licenses
- Certified Soliworkds Associate (CSWA) in Mechanical Design
- Open Water Scuba Diver
Education
- University of Wisconsin Madison | Computer Science | 2023-2025 | Bachelor of Science
- LaGuardia Community College | Computer Science | 2022-2023 | Associate of Science
- National Taiwan Ocean University | Mechanical Engineering | 2019-2021
Support Vector Machine
Project Link
Motivation:
Throughout my journey of learning and applying AI & Machine Learning, I encountered a challenge: the absence of an open-source library in C/C++ for Supervised Learning algorithms. Driven by my passion for Support Vector Machines and the underlying mathematics, I made the decision to embark on the task of building such a library from scratch.
The journey has been both lengthy and enjoyable. I encountered several challenges along the way, each of which I successfully addressed. Despite the algorithm functioning perfectly, I opted to use OpenGL for display, and I am currently focused on refining the accuracy of the display, an aspect that continues to be a work in progress.
As of now, this project is limited to binary classification; however, it has demonstrated an impressive capability by achieving 100% accuracy on linearly separable datasets.
I am committed to further developing this project and aspire to extend its capabilities to handle multiple classifications simultaneously.
Throughout my journey of learning and applying AI & Machine Learning, I encountered a challenge: the absence of an open-source library in C/C++ for Supervised Learning algorithms. Driven by my passion for Support Vector Machines and the underlying mathematics, I made the decision to embark on the task of building such a library from scratch.
The journey has been both lengthy and enjoyable. I encountered several challenges along the way, each of which I successfully addressed. Despite the algorithm functioning perfectly, I opted to use OpenGL for display, and I am currently focused on refining the accuracy of the display, an aspect that continues to be a work in progress.
As of now, this project is limited to binary classification; however, it has demonstrated an impressive capability by achieving 100% accuracy on linearly separable datasets.
I am committed to further developing this project and aspire to extend its capabilities to handle multiple classifications simultaneously.
Phishing Email Filtering
Project Link
Motivation:
During my research collaboration with Dr. Khethavath on phishing email filtering, I have chosen to publish my findings on HuggingFace. The objective is to showcase how Supervised Learning techniques respond to various types of features present in contemporary emails.
I utilize Streamlit as the user interface, enabling users to upload their emails in the .eml file format. Through this interface, they can then interactively explore:
1.What are the features extracted?
2.Whether it's a phishing email or not based on model prediction?
Throughout my research, the models I trained have consistently achieved an accuracy rate exceeding 80%. I am committed to ongoing refinement, ensuring the continued improvement and optimal performance of the models.
During my research collaboration with Dr. Khethavath on phishing email filtering, I have chosen to publish my findings on HuggingFace. The objective is to showcase how Supervised Learning techniques respond to various types of features present in contemporary emails.
I utilize Streamlit as the user interface, enabling users to upload their emails in the .eml file format. Through this interface, they can then interactively explore:
1.What are the features extracted?
2.Whether it's a phishing email or not based on model prediction?
Throughout my research, the models I trained have consistently achieved an accuracy rate exceeding 80%. I am committed to ongoing refinement, ensuring the continued improvement and optimal performance of the models.
LaGuardia Express Search
Project Link | Present Video
I collaborated with fellow students on this project during the LaGCC Hackathon and am proud to share that we secured the 1st prize.
Description: LaGuardia Express Search is a robust tool meticulously crafted to assist LaGuardia Community College students in swiftly accessing the information and resources essential to their needs. Our application provides a seamless experience with features such as text-based search and speech-to-text functionality, catering to users in English, Spanish, and Chinese languages.
Description: LaGuardia Express Search is a robust tool meticulously crafted to assist LaGuardia Community College students in swiftly accessing the information and resources essential to their needs. Our application provides a seamless experience with features such as text-based search and speech-to-text functionality, catering to users in English, Spanish, and Chinese languages.
