Jekyll2024-10-04T16:22:37+00:00https://jpfolch.github.io/feed.xmlJose Pablo FolchI am the chief scientific officer and co-founder of SOLVE chemistry. I am interested in the use of machine learning methods to improve the efficiency and sustainability of chemical processes.Transition Constrained BO via MDPs accepted into NeurIPS 20242024-09-25T11:03:52+00:002024-09-25T11:03:52+00:00https://jpfolch.github.io/seminar/2024/09/25/mdpboNIPS24Very excited that the work I started during my academic placement at ETH has been accepted to NeurIPS 2024! Very grateful to all my collaborators for all the effort put into this amazing paper.

  • Folch, Jose Pablo, Calvin Tsay, Robert M. Lee, Behrang Shafei, Weronika Ormaniec, Andreas Krause, Mark van der Wilk, Ruth Misener, and Mojmír Mutný. “Transition Constrained Bayesian Optimization via Markov Decision Processes”. To appear in Neural Information Processing Systems 37 (2024). Preprint.
]]>I have joined SOLVE Chemistry as co-founder and CSO2024-08-19T11:03:52+00:002024-08-19T11:03:52+00:00https://jpfolch.github.io/seminar/2024/08/19/SOLVE

Excited to announce I have joined SOLVE Chemistry as co-founder and CSO! Founded by Linden Schrecker and backed by Creator Fund, Chemovator, and Imperial College; SOLVE is a company specializing in improving chemical processes through innovative data collection and machine learning approaches. I am looking forward to applying all I researched and learnt during my PhD to help the chemical industry become more efficient.

]]>C&CE Best Paper Award 20232024-06-21T11:03:52+00:002024-06-21T11:03:52+00:00https://jpfolch.github.io/seminar/2024/06/21/CACEBestPaperI am very happy to announce that our paper “Combining Multi-fidelity Modelling and Asynchronous Batch Bayesian Optimisation” has been recognized as the best paper of 2023 by Computers & Chemical Engineering. Imperial College has written an article about it, make sure to check it out.

]]>NeurIPS 20232023-12-15T11:03:52+00:002023-12-15T11:03:52+00:00https://jpfolch.github.io/seminar/2023/12/15/neurIPSI had an amazing time at NeurIPS 2023 in New Orleans. I presented two different workshop papers, the first one at the “Adaptive Experimental Design and Active Learning in the Real World” workshop related to extending SnAKe to more practical settings, including: self-stopping optimization, multi-objective optimization and strict constraints on the movement distances. The second work related to the AirQo collaboration that started at the AI for Social Good seminar back in 2022, Ruby Sedgwick and me recruited an MEng student, Clara Stoddart, whose thesis was accepted into the “Tackling Climate Change with Machine Learning” workshop. The rest of the time I spent it exploring New Orelans, I loved the city and will definitely be visiting again.

Links to the papers can be found here:

  • Folch, Jose Pablo, James Odgers, Shiqiang Zhang, Robert M. Lee, Behrang Shafei, David Walz, Calvin Tsay, Mark van der Wilk, and Ruth Misener. “Practical Path-based Bayesian Optimization”. NeurIPS 2023 Workshop on Adaptive Experimental Design and Active Learning in the Real World. Link.

  • Stoddart, Clara, Lauren Shrack, Richard Sserunjogi, Usman Abdul-Ganiy, Engineer Bainomugisha, Deo Okure, Ruth Misener, Jose Pablo Folch, and Ruby Sedgwick. “Gaussian Processes for Monitoring Air-Quality in Kampala”. NeurIPS 2023 Workshop: Tackling Climate Change with Machine Learning. Link.

]]>Academic Placement at ETH Zurich2023-09-15T11:03:52+00:002023-09-15T11:03:52+00:00https://jpfolch.github.io/seminar/2023/09/15/zurichETHI have just returned from finishing an academic placement at ETH Zurich, at the LAS research group led by Andreas Krause. During my time there I had the privilage of working closely with Mojmír Mutny. It was an extremely rewarding experience, and I hope everyone will see the amazing work we did together as a publication soon. Zurich is now one of my favourities cities, surrounded by beautiful nature, and with the most amazing food. Find below a few pictures I took during a weekend hike!

]]>BayesComp 20232023-03-20T11:03:52+00:002023-03-20T11:03:52+00:00https://jpfolch.github.io/seminar/2023/03/20/bayescomp22I recently return from Levi, Finland, where I attended BayesComp 2023. The conference explored the latest Bayesian computational methods, such as MCMC and variational Bayes. It was very fun to get to learn about the latest research, see how it is applied to the real world, and to catch up with some friends. The conference is also colloquially known as MCMSki, as the conference centre was in the middle of a ski resort! During my time there, I took the opportunity to learn to snowboard and to see the aurora borealis.

]]>New Computers & Chemical Engineering Publication2023-02-20T11:03:52+00:002023-02-20T11:03:52+00:00https://jpfolch.github.io/seminar/2023/02/20/mfboomI am pleased to announce that our latest paper, “Combining Multi-Fidelity Modelling and Asynchronous Batch Bayesian Optimization” has been accepted into the latest issue of the journal Computers & Chemical Engineering. The link to the publication can be found here. We also made videos detailing the work, both a short and a long version.

]]>AIChE Conference 2022 Talk2022-11-18T11:03:52+00:002022-11-18T11:03:52+00:00https://jpfolch.github.io/seminar/2022/11/18/aicheI was pleased to present SnAKe at the annual AIChE (American Institute of Chemical Engineers) conference, in Phoenix, AZ, USA. There were a lot of interesting talks about combining Chemical Engineering with Machine Learning, and I was able to meet many interesting people! I am looking forward to attend more conferences in the near future.

]]>SnAKe accepted into NeurIPS 20222022-09-15T11:03:52+00:002022-09-15T11:03:52+00:00https://jpfolch.github.io/seminar/2022/09/15/snake-acceptI am pleased to announce that my very first paper, SnAKe, has been accepted into NeurIPS 2022! Stay tuned for a video presentation and poster session in New Orleans!

]]>Bocconi Summer School in Lake Como2022-07-25T16:03:52+00:002022-07-25T16:03:52+00:00https://jpfolch.github.io/seminar/2022/07/25/bocconi-ssI recently attended the almost two week summer school organized by Bocconi, in collaboration with the StatML CDT. It was one of the best experiences in my PhD so far! The official name for the 2022 edition was “Random structures and combinational statistics”, or as it was described to us on the first day, “Random stuff”. The first week was taught by Gabor Lugosi, and it related to concentration inequalities, and by the end we we had an introduction to random graphs and the hidden clique problem. The second week was taught by Luc Devroye, where we explored random graphs and trees in far more detail. It was more theoretical than what I am usually used to in my PhD, but it was great fun nonetheless! I particularly enjoyed that it reminded me of my undergraduate degree, but this time the class room was a lot more beautiful:

In addition to the lectures, it was very enjoyable to spend two weeks living around Lake Como! The everyday walk from the hotel to the Villa was particularly beautiful (even if, at times, it was too hot!). We spend most of our evenings exploring the city, trying all types of Italian food, and spending time with the other amazing students from the summer school, I think we all made life-time friends! Highly recommend attending if you ever have the opportunity!

]]>