<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Deep Learning on Zitao Liao</title><link>http://lzteddy.com/tags/deep-learning/</link><description>Recent content in Deep Learning on Zitao Liao</description><generator>Hugo -- gohugo.io</generator><language>en-us</language><lastBuildDate>Sun, 19 Apr 2026 00:00:00 +0000</lastBuildDate><atom:link href="http://lzteddy.com/tags/deep-learning/index.xml" rel="self" type="application/rss+xml"/><item><title>Research Experience</title><link>http://lzteddy.com/p/research-experience/</link><pubDate>Sun, 19 Apr 2026 00:00:00 +0000</pubDate><guid>http://lzteddy.com/p/research-experience/</guid><description>&lt;h2 id="competition-on-llm-designed-ea-gecco"&gt;Competition on LLM Designed EA, GECCO
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Mar 2026&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Collaborated in a team to design adaptive multi-operator evolutionary algorithms using the LLM4AD platform and Evolution of Heuristics (EoH).&lt;/li&gt;
&lt;li&gt;Designed a module based on the Multi-Armed Bandit (MAB) algorithm for adaptive operator selection during optimization.&lt;/li&gt;
&lt;li&gt;Achieved optimal results on &lt;strong&gt;17/24&lt;/strong&gt; GNBG benchmark problems.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="research-on-combinatorial-optimization-methods-based-on-deep-learning-and-development-of-the-easyco-platform"&gt;Research on Combinatorial Optimization Methods Based on Deep Learning and Development of the EasyCO Platform
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Sep 2025 - Present&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Explored graph neural network based data representations and feature learning mechanisms integrated with reinforcement learning frameworks (for example REINFORCE).&lt;/li&gt;
&lt;li&gt;Contributed to the open-source EasyCO platform for building general-purpose efficient solvers for industrial applications.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="autonomous-driving-video-annotation-system-based-on-videollama2"&gt;Autonomous Driving Video Annotation System Based on VideoLLaMA2
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Apr 2025 - Jun 2025&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Designed English prompts focused on key driving elements to generate descriptions centered on ego-vehicle behavior and interactions with preceding vehicles.&lt;/li&gt;
&lt;li&gt;Fine-tuned the model to optimize accuracy.&lt;/li&gt;
&lt;/ul&gt;</description></item><item><title>Projects</title><link>http://lzteddy.com/p/projects/</link><pubDate>Sat, 18 Apr 2026 00:00:00 +0000</pubDate><guid>http://lzteddy.com/p/projects/</guid><description>&lt;h2 id="intelligent-sanitation-robot-implementation-based-on-deep-learning"&gt;Intelligent Sanitation Robot Implementation Based on Deep Learning
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Jul 2025&lt;br&gt;
&lt;strong&gt;Note:&lt;/strong&gt; Main practical assignment for the NUS SoC Summer Workshop Program.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Applied transfer learning on YOLOv7 for multi-object detection: trash (paper and bottles), cigarettes, violent behavior, and human falls.&lt;/li&gt;
&lt;li&gt;Built a crane-style structure with servo-controlled grabbing arms and launching mechanism on a robot car.&lt;/li&gt;
&lt;li&gt;Integrated Arduino and Raspberry Pi for hardware control.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="real-time-facial-emotion-analysis-system-on-jetson-nano"&gt;Real-time Facial Emotion Analysis System on Jetson Nano
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Dec 2025&lt;br&gt;
&lt;strong&gt;Note:&lt;/strong&gt; Main practical assignment for the Deep Learning course.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Fine-tuned a lightweight face detection model (CNN with RFB modules) on a custom emotion dataset.&lt;/li&gt;
&lt;li&gt;Deployed models on the Jetson Nano edge computing platform.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="modeling-and-optimization-of-voting-mechanisms-in-dancing-with-the-stars"&gt;Modeling and Optimization of Voting Mechanisms in Dancing with the Stars
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Feb 2026&lt;br&gt;
&lt;strong&gt;Note:&lt;/strong&gt; Achievement of 2026 MCM.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Addressed voting controversies using Bayesian inverse estimation, counterfactual simulations, and XGBoost with the CPISeq framework.&lt;/li&gt;
&lt;li&gt;Proposed a Four-Strike Rank Fusion voting rule for fairness in high-conflict scenarios.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="pipelined-risc-v-cpu-design-and-implementation-based-on-verilog"&gt;Pipelined RISC-V CPU Design and Implementation Based on Verilog
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; May 2025&lt;br&gt;
&lt;strong&gt;Note:&lt;/strong&gt; Main practical assignment for the Computer Organization course.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Implemented a five-stage pipelined CPU (RISC-V 32I) on the EGO1 FPGA board.&lt;/li&gt;
&lt;li&gt;Added UART communication, branch prediction, and a Python GUI for real-time register and pipeline visualization.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="performance-comparison-and-evaluation-of-opengauss-vs-postgresql"&gt;Performance Comparison and Evaluation of openGauss vs. PostgreSQL
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Dec 2024&lt;br&gt;
&lt;strong&gt;Note:&lt;/strong&gt; Main practical assignment for the Principles of Database Systems (H) course.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Compared openGauss and PostgreSQL via pgbench across query performance, concurrency, connection efficiency, import speed, resource utilization, and security.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="java-swing-based-match-3-puzzle-game"&gt;Java Swing-Based Match-3 Puzzle Game
&lt;/h2&gt;&lt;p&gt;&lt;strong&gt;Time:&lt;/strong&gt; Dec 2023&lt;br&gt;
&lt;strong&gt;Note:&lt;/strong&gt; Main practical assignment for the Introduction to Computer Programming A (H) course.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Designed and implemented a Match-3 game with multi-level progression, score and move limits, and both manual and auto-play modes.&lt;/li&gt;
&lt;/ul&gt;</description></item></channel></rss>