<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Multimodal Learning | Yik Yu Ng</title><link>https://yyn-anson.github.io/tags/multimodal-learning/</link><atom:link href="https://yyn-anson.github.io/tags/multimodal-learning/index.xml" rel="self" type="application/rss+xml"/><description>Multimodal Learning</description><generator>Hugo Blox Builder (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Sun, 31 May 2026 00:00:00 +0000</lastBuildDate><image><url>https://yyn-anson.github.io/media/icon_hu7729264130191091259.png</url><title>Multimodal Learning</title><link>https://yyn-anson.github.io/tags/multimodal-learning/</link></image><item><title>Privacy-Preserving Multimodal AI for MS Prognosis</title><link>https://yyn-anson.github.io/project/multimodal-ms-prognosis/</link><pubDate>Sun, 31 May 2026 00:00:00 +0000</pubDate><guid>https://yyn-anson.github.io/project/multimodal-ms-prognosis/</guid><description>&lt;h2 id="overview">Overview&lt;/h2>
&lt;p>This research develops a privacy-preserving multimodal large language model system for personalized multiple-sclerosis lesion-activity prognosis. The system combines clinical narratives and MRI while keeping model inference local.&lt;/p>
&lt;h2 id="research-direction">Research direction&lt;/h2>
&lt;ul>
&lt;li>A retrieval-synthesis-critique workflow assembles grounded evidence for the final assessment.&lt;/li>
&lt;li>The design emphasizes privacy, traceability, and cautious use in a clinical decision-support setting.&lt;/li>
&lt;li>Current work studies systematic failure subgroups through multimodal decision embeddings.&lt;/li>
&lt;/ul>
&lt;p>This project is conducted at Mila - Quebec AI Institute under the supervision of Professor Tal Arbel.&lt;/p></description></item></channel></rss>