CSML IIT Lab
@pontilgroup.bsky.social
Computational Statistics and Machine Learning (CSML) Lab | PI: Massimiliano Pontil | Webpage: csml.iit.it | Active research lines: Learning theory, ML for dynamical systems, ML for science, and optimization.
11/ One of our most exciting results:
[P8] NeurIPS 2024 proposed Neural Conditional Probability (NCP) to efficiently learn conditional distributions. It simplifies uncertainty quantification and guarantees accuracy for nonlinear, high-dimensional data.
[P8] NeurIPS 2024 proposed Neural Conditional Probability (NCP) to efficiently learn conditional distributions. It simplifies uncertainty quantification and guarantees accuracy for nonlinear, high-dimensional data.
January 15, 2025 at 2:34 PM
11/ One of our most exciting results:
[P8] NeurIPS 2024 proposed Neural Conditional Probability (NCP) to efficiently learn conditional distributions. It simplifies uncertainty quantification and guarantees accuracy for nonlinear, high-dimensional data.
[P8] NeurIPS 2024 proposed Neural Conditional Probability (NCP) to efficiently learn conditional distributions. It simplifies uncertainty quantification and guarantees accuracy for nonlinear, high-dimensional data.
7/ 📈 Scaling up:
[P4] NeurIPS 2023 introduced a Nyström sketching-based method to reduce computational costs from cubic to almost linear without sacrificing accuracy. Validated on massive datasets like molecular dynamics, see figure.
[P4] NeurIPS 2023 introduced a Nyström sketching-based method to reduce computational costs from cubic to almost linear without sacrificing accuracy. Validated on massive datasets like molecular dynamics, see figure.
January 15, 2025 at 2:34 PM
7/ 📈 Scaling up:
[P4] NeurIPS 2023 introduced a Nyström sketching-based method to reduce computational costs from cubic to almost linear without sacrificing accuracy. Validated on massive datasets like molecular dynamics, see figure.
[P4] NeurIPS 2023 introduced a Nyström sketching-based method to reduce computational costs from cubic to almost linear without sacrificing accuracy. Validated on massive datasets like molecular dynamics, see figure.
6/ [P3] ICML 2024 addressed a critical issue in TO-based modeling: reliable long-term predictions.
Our Deflate-Learn-Inflate (DLI) paradigm ensures uniform error bounds, even for infinite time horizons. This method stabilized predictions in real-world tasks; see the figure.
Our Deflate-Learn-Inflate (DLI) paradigm ensures uniform error bounds, even for infinite time horizons. This method stabilized predictions in real-world tasks; see the figure.
January 15, 2025 at 2:34 PM
6/ [P3] ICML 2024 addressed a critical issue in TO-based modeling: reliable long-term predictions.
Our Deflate-Learn-Inflate (DLI) paradigm ensures uniform error bounds, even for infinite time horizons. This method stabilized predictions in real-world tasks; see the figure.
Our Deflate-Learn-Inflate (DLI) paradigm ensures uniform error bounds, even for infinite time horizons. This method stabilized predictions in real-world tasks; see the figure.
4/ 🔑 The journey began with:
[P1] NeurIPS 2022
We introduced the first ML formulation for learning TO, which led to the development of the open-source Kooplearn library. This step laid the groundwork for exploring the theoretical limits of operator learning from finite data.
[P1] NeurIPS 2022
We introduced the first ML formulation for learning TO, which led to the development of the open-source Kooplearn library. This step laid the groundwork for exploring the theoretical limits of operator learning from finite data.
January 15, 2025 at 2:34 PM
4/ 🔑 The journey began with:
[P1] NeurIPS 2022
We introduced the first ML formulation for learning TO, which led to the development of the open-source Kooplearn library. This step laid the groundwork for exploring the theoretical limits of operator learning from finite data.
[P1] NeurIPS 2022
We introduced the first ML formulation for learning TO, which led to the development of the open-source Kooplearn library. This step laid the groundwork for exploring the theoretical limits of operator learning from finite data.