PsiQuantum’s Post

Pete Shadbolt, cofounder and chief scientific officer of PsiQuantum at #EmTechMIT on why traditional #computing will eventually reach a ceiling: “Machine learning has been very successful in learning from #data and making good approximations in areas where we have large volumes of data. However, there are problems that we would like to solve from first principles or in areas where we don’t have good access to data. So drug design, #materials design, partial differential equation solving… these are the types of problems that we resort to #supercomputers for. Many of those problems have the property that they scale exponentially badly as you increase the size of the molecule or the size of the area of material that you’re trying to simulate. And so for some of those problems, we don’t know of any classical algorithm that would solve those problems in any kind of acceptable time, even with a conventional #computer the size of Manhattan. So there are a set of problems that we don’t expect to ever solve with conventional supercomputers. And that’s why people are going to these great lengths and why there is so much excitement about #quantum computing.”

This past summer, my amazing co-editor Ruda Zhang and I had the privilege of editing a special issue on Advancements and Applications of Multifidelity Machine Learning in Engineering for the Journal of Machine Learning for Modeling and Computing. Here’s a brief introduction to the five papers published in this special issue: "Multi-fidelity Machine Learning for Uncertainty Quantification and Optimization" – In this editorial perspective, we provide an overview of some of the state-of-the-art techniques in multifidelity modeling for UQ and optimization, highlighting the critical gaps and opportunities therein. "Mesh Refinement as Probabilistic Learning" – Authored by Philippe Hawi and Roger Ghanem, this paper explores how simultaneously observing low-fidelity and high-fidelity FEM solutions allows for learning the joint probability distribution of the quantity of interest. "Robust Topology Optimization Using Multi-fidelity Variational Autoencoders" – Authored by Rini Gladstone, Mohammad Amin Nabian, Vahid Keshavarzzadeh, and Hadi Meidani, this work accelerates robust topology optimization by building a low-dimensional search space and using multi-fidelity VAE surrogates. "Design of Experiments via Multi-Fidelity Surrogates and Statistical Sensitivity Measures" – Authored by David J. Gillcrist, Mazdak Tootkaboni and Yanlai Chen, this paper presents a novel multi-fidelity framework for identifying the optimal experimental setup, from a finite set of possibilities, for precise parameter estimation. "Hybrid Chance-Constrained Optimal Power Flow Under Load and Renewable Generation Uncertainty Using Enhanced Multi-Fidelity Graph Neural Networks" – Authored by Kamiar Khayambashi, Md Abul Hasnat, and me, this paper investigates how a multi-fidelity GNN surrogate can assist in solving the optimal power flow problem under high-dimensional uncertainty. Check out these fantastic papers, and feel free to reach out to us or the authors if you have any questions! Special thanks to the editor-in-chief, Dongbin Xiu, for this incredible opportunity. Below is the link to the Special Issue: https://lnkd.in/gaM3dDbG

HQS attends DevCamp24 at CyberForum in Karlsruhe!🎉 Interested in topics like quantum computing, AI/machine learning, and IT security? 💻 Join us for our presentation at DevCamp on October 29 at CyberForum. 📅 In our talk on "Roqoqo, a quantum computing toolkit in Rust," you'll learn more about Roqoqo's capabilities and gain insights into quantum programming, measurement execution, and serialization. Here are some features of our open-source software Roqoqo: 🔶 A comprehensive toolkit for representing quantum programs, including circuits and measurement information. 🔶 A lightweight runtime for performing quantum measurements. Serialization capabilities for quantum circuits and measurement data. 🔶 Optional interfaces for seamless integration with devices, simulators and toolkits (e.g. qoqo_quest, qoqo_mock, qoqo_qasm). ➡️ Explore more: https://lnkd.in/ePudmtWX 👉 For registration and further information about the event please click here: https://lnkd.in/drTMetr6 We look forward to seeing you! ✌ #karlsruhe #ai #machinelearning #quantencomputing #itsecurity #barcamp

This week, Google Research released a paper on arXiv showing that boosting can now optimize any loss function without relying on first-order information. This breakthrough opens up new possibilities for machine learning models. Key Highlights: - Versatility: The new boosting algorithm can handle non-convex, non-differentiable, and even discontinuous loss functions. - Quantum Calculus: Utilizes quantum calculus to avoid derivatives, making it more efficient. - Zeroth-Order Optimization: Extends the principles of gradient-based optimization to zeroth-order, using only loss values. Curious about the details? Check out the full paper here: https://lnkd.in/drSjaZhr #MachineLearning #Boosting #AIResearch #QuantumCalculus #ZerothOrderOptimization

CECAM-Lorentz call 2025-2026 is now open! Would you like to organize your own computational science workshop? Send us your proposal before 13 November 2024. Although all topics are welcome, in this year’s call, we aim to specifically encourage proposals addressing the question of sustainable computing. How can scientific software be made more energy-efficient? Possible topics include, but are not limited to, sustainable codes for atomistic and molecular simulations, data science, scientific image analysis, or sustainable ML/AI. How to join See the call webpage for more information about the procedure and how to join: https://lnkd.in/eqP8Zf34

Excited to have presented my paper on 'Recurrent Neural Network Efficient Urban Traffic Management Through Integration of Machine Learning & Edge Computing' at the International Conference on Intelligent Computing and Technology. Grateful for the chance to exchange ideas with industry experts. #IntelligentComputing #MachineLearning #EdgeComputing #TrafficManagement

Consensus learning: A novel decentralised ensemble learning paradigm - a new paper from Flare Network | Flare Research The widespread adoption of large-scale #machinelearning models in recent years highlights the need for distributed computing for efficiency and scalability. This work introduces a novel distributed machine learning paradigm – #consensuslearning – which combines classical ensemble methods with #consensus #protocols deployed in peer-to-peer systems. These algorithms consist of two phases: first, participants develop their models and submit predictions for any new data inputs; second, the individual predictions are used as inputs for a communication phase, which is governed by a consensus protocol. Consensus learning ensures user data privacy, while also inheriting the safety measures against Byzantine attacks from the underlying consensus mechanism. The authors provide a detailed theoretical analysis for a particular consensus protocol and compare the performance of the consensus learning ensemble with centralised ensemble learning algorithms. The discussion is supplemented by various numerical simulations, which describe the robustness of the algorithms against Byzantine participants. 

🗣️ Tomorrow, at 13:00, we will enjoy the talk by one of our PhD Students, Aristide Tanyi-Jong Akem in our #NetworksWeeklySeminar. 🔹 During the seminar, he will address how ML inference in programmable switches can enable accelerated attack detection and mitigation in Smart Grids (SGs) at line rate with a sub-microsecond delay. 🔸 What is an ML inference?👉 it’s the process of running live data points into a machine learning algorithm (or “ML model”) to calculate an output such as a single numerical score. 👉 All the details about the event: https://lnkd.in/dBj52k62

#Tech Everything You See Is a Computational Process, If You Know How to Look: Computer scientist Lance Fortnow writes that by embracing the computations that surround us, we can begin to understand and tame our seemingly random world. #ArtificialIntelligence #MachineLearning #DataScience

The talk about discovering QEC codes with RL that I gave last December at The Alan Turing Institute's Physics-informed machine learning meets engineering seminar series is now available on Youtube. Check it out in case you're interested! 😊 https://lnkd.in/dgq_8JrW