<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Vinamr Jain</title><description>Machine learning, Cornell</description><link>https://vinamr.dev/</link><item><title>Bayesian Decision Making in Unknown Dynamics</title><link>https://vinamr.dev/papers/decision-making-unknown-dynamics/</link><guid isPermaLink="true">https://vinamr.dev/papers/decision-making-unknown-dynamics/</guid><description>A weather balloon 🎈 cannot steer itself. It can only go up or down ↕️, and the wind 🌬️ at each height carries it somewhere different. To reach a target 🎯 it has to ride the tide 🌊, but it doesn&apos;t know where the flow is going! (well maybe it pays attention to weather forecasts 📡 before going out): it only knows the wind where it has already been. So it has two jobs at once: learn the wind, and get to where it is going. Should it take a detour to learn more (explore 🧭)? or head for the target on a shaky guess (exploit 🎲)?</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>zermelo</title><link>https://vinamr.dev/builds/zermelo/</link><guid isPermaLink="true">https://vinamr.dev/builds/zermelo/</guid><description>zermelo is a simulation environment 🕹️ that models autonomous systems like a weather balloon 🎈 navigating in the wind 🌬️, or a saildrone ⛵ powered by free will 🧠. It is the environment behind Bayesian Decision Making in Unknown Dynamics. Repository name inspired from Zermelo&apos;s navigation problem.</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Thompson Sampling using Prior-fitted Diffusion Transformers</title><link>https://vinamr.dev/papers/pfn-thompson-sampling/</link><guid isPermaLink="true">https://vinamr.dev/papers/pfn-thompson-sampling/</guid><description>## The problem Imagine you work in a startup and are responsible for running experiments to design a new superconducting material. But the investors are in a hurry! ⏳ So you only get a handful of attempts. Is there a better strategy than randomly guessing suitable designs? ## Explore or exploit? The range of conductivity values of different design possibilities varies continuously. You may have some domain knowledge, for eg. similar materials have similar properties. (Is there a way you can use this prior knowledge or belief about material properties?) Suppose you&apos;ve already designed a few materials and you need to decide what to design next. Maybe you should design a material similar to the best one so far, but maybe if you design one completely different from the rest you might be pleasantly surprised (High risk = High reward?). This is known as the exploration-exploitation tradeoff in decision problems. Since you don&apos;t know anything about the landscape of material properties (except that maybe it&apos;s a continuous function), this particular kind of decision problem is known as black-box optimization. If only someone had been in a similar situation 🥹 so you could take inspiration from what they did! ## Practice makes perfect It is now you remember that god is within you, so using your domain knowledge: 1. You create a simulator on your computer that generates synthetic &quot;worlds&quot; or function landscapes of material properties, similar to the ones observed in the real world 2. In these synthetic worlds you similarly run a bunch of experiments 3. Based on those experiments, you ask &quot;Given the designed materials and their observed properties, what design would have the highest conductivity?&quot;, only this time, since you created this world, you know the answer! Wait, why is this a good idea? Assuming you&apos;re not living under a rock, you&apos;ve heard the average tech-bro talk about training AI models: Using a bunch of input-output examples you can get a model which, given an input, gives a prediction of the output! In fact, you can use this simulator to generate synthetic datasets, where you input a bunch of experiment results as context and you want to predict the likelihood of a candidate being the best design (aka the posterior belief). The AI models (aka artificial neural networks) fitted (trained) using synthetic data generated using your prior (domain) knowledge are known as Prior-fitted Networks! ## Test what&apos;s best The flavour of decision rules (aka acquisition rules) that uses a likelihood or probability distribution to model the uncertainty to perform an &quot;optimization&quot; task, is called Bayesian Optimization (Remember Bayesian probability? you&apos;d remember if you didn&apos;t skip high school 😏). A popular acquisition rule is to pick the next candidate design with probability proportional to the posterior belief of it being the best candidate. This rule is known as Thompson Sampling 🎲. Finally how do we get these posterior belief values over a continuous range of design possibilities? Diffusion Transformers are AI models that work by diffusing or transporting potential design points randomly drawn from noise to more concentrated regions proportional to the posterior belief to get a continuous distribution. ## All together now Bringing all these parts together, this work developed Thompson Sampling using Prior-fitted Diffusion Transformers for black-box optimization! (P.S. Random search lowkey sucks!)</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Higher-Order Quantum Reservoir Computing for Non-Intrusive Reduced-Order Models</title><link>https://doi.org/10.2514/6.2026-1419</link><guid isPermaLink="true">https://doi.org/10.2514/6.2026-1419</guid><description>Quantum hardware inspired machine learning algorithm for sea surface temperature forecasting.</description><pubDate>Thu, 08 Jan 2026 00:00:00 GMT</pubDate></item><item><title>Machine Learning Pipelines for the Design of Solid-State Electrolytes</title><link>https://doi.org/10.1039/D5MH01525A</link><guid isPermaLink="true">https://doi.org/10.1039/D5MH01525A</guid><description>Review, perspectives and forward looking recommendations on the state of machine learning research for battery material design.</description><pubDate>Thu, 01 Jan 2026 00:00:00 GMT</pubDate></item><item><title>AI-Driven Rational Design of Promiscuous and Selective Plastic-Binding Peptides</title><link>https://doi.org/10.1039/D5SC04903B</link><guid isPermaLink="true">https://doi.org/10.1039/D5SC04903B</guid><description>AI model paired with discrete optimization to design peptides that bind to multiple/selectively to different microplastics.</description><pubDate>Wed, 01 Oct 2025 00:00:00 GMT</pubDate></item></channel></rss>