https://ines2r.github.io/tags/ai/ Recent content in AI on Inès de Rosnay Hugo -- gohugo.io en-us Tue, 20 Jan 2026 00:00:00 +0000 https://ines2r.github.io/p/welcome/ Tue, 20 Jan 2026 00:00:00 +0000 https://ines2r.github.io/p/welcome/ <img src="https://ines2r.github.io/p/welcome/cover.jpg" alt="Featured image of post Working Notes" /><style> /* 2. LE TABLEAU (Général) */ .article-content table { width: 100% !important; display: table !important; border-collapse: collapse; margin-bottom: 2em; } /* Par défaut, on laisse les colonnes s'ajuster (pour le tableau des couleurs) */ .article-content td { vertical-align: top !important; padding: 10px !important; border-bottom: 1px solid #eee; } /* 3. EXCEPTION POUR LES GRAPHES (Législatures) */ /* Si le tableau a 3 colonnes, on force les largeurs égales pour vos graphes */ .article-content table tr th:first-child:nth-last-child(3), .article-content table tr th:first-child:nth-last-child(3) ~ th, .article-content table tr td:first-child:nth-last-child(3), .article-content table tr td:first-child:nth-last-child(3) ~ td { width: 33.33% !important; } /* 4. GESTION INTELLIGENTE DES IMAGES */ .article-content table img { height: auto !important; display: block; } /* Si l'image est un "badge" de couleur (Shields.io), on la garde petite */ .article-content table img[src*="shields.io"] { width: 90px !important; /* Taille fixe pour vos carrés de couleur */ display: inline-block; } /* Si c'est un graphique (pas un badge), il prend toute la place de sa colonne */ .article-content table img:not([src*="shields.io"]) { width: 100% !important; max-width: none !important; } </style> <hr> <h1 id="working-notes">Working Notes </h1><p><strong>Exploring AI, Machine Learning, and Graphs — with Limited Resources</strong></p> <p>This blog is a technical notebook to explore topics I’m curious about and bridge the gap between theory and practice through hands-on experiments. Everything here—code, data, and analysis—is hosted on GitHub to ensure others can inspect, reproduce, or build upon my work.</p> <hr> <h3 id="current-interests--research-ideas">Current Interests &amp; Research Ideas </h3><p>These are topics I am thinking about:</p> <ul> <li><strong>AI &amp; Machine Learning:</strong> LLM fine-tuning, data poisoning, and predictive modeling (academic challenges/real-world data).</li> <li><strong>Graph Theory &amp; Network Science:</strong> Analyzing political networks (parliamentary voting behavior) and modeling epidemic spreads.</li> <li><strong>Quantum Computing &amp; Cryptography:</strong> Exploring the fundamentals of quantum information—such as running single-qubit circuits via the Felis framework and Alice &amp; Bob’s cat qubit technology.</li> <li><strong>Decision Systems &amp; Strategy:</strong> Game theory applied to sports (tennis) or board games, and reinforcement learning.</li> <li><strong>Reflections on AI &amp; Sovereignty:</strong> Analyzing how Large Language Models shape public opinion and impact democratic stability. I’m particularly interested in the dominance of a few tech giants and how states leverage these tools to project influence and challenge national autonomy.</li> </ul> <hr> <h3 id="the-low-tech-constraint">The &ldquo;Low-Tech&rdquo; Constraint </h3><p>I run my experiments on a <strong>2017 MacBook Air (8GB RAM, Dual-Core i5)</strong> without paid cloud services. This constraint encourages me to focus on small LLMs and lightweight techniques like LoRA. For me, it’s about deep understanding and efficiency rather than massive scale.</p> <hr> <h2 id="closing-note">Closing Note </h2><p>This is a space for curiosity and open-source collaboration. Everything shared here is open for you to reuse, adapt, or critique. If a project sparks your interest, feel free to dive in and share your findings.</p> <blockquote> <p>Photo by <a class="link" href="https://unsplash.com/@pawel_czerwinski" target="_blank" rel="noopener" >Pawel Czerwinski</a> on <a class="link" href="https://unsplash.com/" target="_blank" rel="noopener" >Unsplash</a></p> </blockquote>