<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Recent &amp; Upcoming Talks | Sara Candussio</title><link>https://gaoithee.github.io/saracandussio.github.io/event/</link><atom:link href="https://gaoithee.github.io/saracandussio.github.io/event/index.xml" rel="self" type="application/rss+xml"/><description>Recent &amp; Upcoming Talks</description><generator>Hugo Blox Builder (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Fri, 24 Apr 2026 12:05:00 +0000</lastBuildDate><image><url>https://gaoithee.github.io/saracandussio.github.io/media/icon_hu7729264130191091259.png</url><title>Recent &amp; Upcoming Talks</title><link>https://gaoithee.github.io/saracandussio.github.io/event/</link></image><item><title>Reading Between the Tokens: Uncovering the Semantic Minima of AI Monologues</title><link>https://gaoithee.github.io/saracandussio.github.io/event/clcg-linguistics-lunch/</link><pubDate>Fri, 24 Apr 2026 12:05:00 +0000</pubDate><guid>https://gaoithee.github.io/saracandussio.github.io/event/clcg-linguistics-lunch/</guid><description>&lt;p>Invited talk at the &lt;a href="https://www.rug.nl/research/clcg/colloquia_discussiongroups_linguisticevents/linguistics_lunch/" target="_blank" rel="noopener">CLCG Linguistics Lunch&lt;/a> at the University of Groningen.&lt;/p>
&lt;p>Chain-of-Thought prompting asks models to reason step by step — but most of what they write is filler. This talk presents work on identifying the semantic minima of AI reasoning: the tiny subset of tokens that actually carry predictive weight, detectable in real time from the model&amp;rsquo;s internal states. Erasing up to 95% of the output leaves a sparse set of words that still perfectly predicts the correct answer.&lt;/p></description></item><item><title>Large Language Models: Potenzialità, Limiti e Sistemi Multi-Agent</title><link>https://gaoithee.github.io/saracandussio.github.io/event/novalia-llm-workshop/</link><pubDate>Mon, 15 Dec 2025 00:00:00 +0000</pubDate><guid>https://gaoithee.github.io/saracandussio.github.io/event/novalia-llm-workshop/</guid><description>&lt;figure>&lt;img src="novalia.png"
alt="Workshop at Novalia, Trieste — December 2025.">&lt;figcaption>
&lt;p>Workshop at Novalia, Trieste — December 2025.&lt;/p>
&lt;/figcaption>
&lt;/figure>
&lt;p>Part of a two-session seminar series on digital transformation, co-organized with &lt;a href="https://www.ip4fvg.it/" target="_blank" rel="noopener">IP4FVG&lt;/a> and the University of Trieste.&lt;/p>
&lt;p>This session covered the inner workings of Large Language Models, multi-agent architectures, and fine-tuning strategies — with an eye toward practical business applications and a frank discussion of current limitations.&lt;/p>
&lt;p>The core message: the challenge isn&amp;rsquo;t just adopting AI, but integrating it strategically to augment rather than replace human potential.&lt;/p></description></item><item><title>Bridging Logic and Learning: Decoding Temporal Logic Embeddings via Transformers</title><link>https://gaoithee.github.io/saracandussio.github.io/event/dagstuhl-rtg-symposium/</link><pubDate>Mon, 24 Nov 2025 00:00:00 +0000</pubDate><guid>https://gaoithee.github.io/saracandussio.github.io/event/dagstuhl-rtg-symposium/</guid><description>&lt;p>Poster presentation of our ECML-PKDD 2025 paper at the RTG Symposium, held at the legendary &lt;a href="https://www.dagstuhl.de/" target="_blank" rel="noopener">Schloss Dagstuhl&lt;/a> — a week-long gathering of PhD researchers from Max Planck Institute and Saarland University, with guest talks by &lt;a href="https://mega.seas.harvard.edu/" target="_blank" rel="noopener">Mor Geva Pipek&lt;/a> and &lt;a href="https://ai-lab.units.it/?page_id=139" target="_blank" rel="noopener">Luca Bortolussi&lt;/a>.&lt;/p></description></item></channel></rss>