Maryam Najafian, PhD

I’m truly honored to be invited by the IEEE BioSensors 2026 organizing team to deliver a tutorial on: Practical AI/ML for Biosignals: Bridging Statistical Features and Multi-Task Learning Biosignals are rich, messy, and highly variable. In this tutorial, we offer a practical entry point into AI/ML for biosensing by starting from what makes signals like electroencephalography (EEG) and photoplethysmography (PPG) different from typical datasets: their frequency–magnitude structure, noise and artifacts, and strong variability across people and hardware (often experienced as domain shift). We show how this domain knowledge should shape the way you represent signals and choose models. We then use statistical analysis as the bridge from raw waveforms to meaningful features, connecting intuitive signal properties to time-domain statistics and frequency-domain summaries that often carry the most useful information. From there, we explain how classical machine learning uses these engineered features, how deep learning can learn representations directly from raw inputs, and how hybrid pipelines can combine both worlds. We also discuss practical, lightweight approaches to interpretability and explainability that help researchers and practitioners trust and debug biosignal models. Finally, we focus on multi-task learning as an extension for many biosensing problems. We cover how to define tasks that help rather than fight each other, why multi-loss training can become unstable, and how adaptive loss weighting can reduce manual tuning and keep training balanced. Participants will leave with concrete pitfalls to avoid and practical takeaways they can apply immediately.

23

🚀 𝗖𝗙𝗗 𝗯𝗲𝘆𝗼𝗻𝗱 𝗺𝗲𝘀𝗵𝗲𝘀: 𝗧𝗨 𝗗𝗮𝗿𝗺𝘀𝘁𝗮𝗱𝘁 𝗮𝗻𝗱 𝗜𝗔𝗡𝗨𝗦 𝗦𝗶𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻 𝗴𝗼 𝗺𝗲𝘀𝗵𝗹𝗲𝘀𝘀! As part of the #NHR4CES initiative (National High Performance Computing for Computational Engineering Science), a remarkable milestone has been achieved in the 𝗲𝘃𝗼𝗹𝘂𝘁𝗶𝗼𝗻 𝗼𝗳 𝗖𝗙𝗗 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸𝘀. M. Elwardi Fadeli has introduced a 𝗻𝗼𝘃𝗲𝗹 𝗰𝗼𝗱𝗲 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 that redefines how we approach 𝗺𝗲𝘀𝗵𝗹𝗲𝘀𝘀 𝗺𝗲𝘁𝗵𝗼𝗱𝘀. His work - presented at the Math-to-Product Conference 2025 in Valencia - lays the groundwork for a 𝗻𝗲𝘄 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝗺𝗼𝗱𝘂𝗹𝗮𝗿, 𝗵𝘆𝗯𝗿𝗶𝗱, 𝗮𝗻𝗱 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲-𝗮𝗴𝗻𝗼𝘀𝘁𝗶𝗰 𝘀𝗶𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻 𝗲𝗰𝗼𝘀𝘆𝘀𝘁𝗲𝗺𝘀. At the heart of this framework is a 𝗰𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁-𝘁𝗼-𝗰𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁 𝗶𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲 𝗽𝗿𝗼𝘁𝗼𝗰𝗼𝗹 which has been inspired by the Language Server Protocol in modern IDEs. This design paradigm enables: ➡️ Seamless hybridization of meshless and mesh-based solvers ➡️ Dynamic negotiation of capabilities between independently developed components ➡️ Rapid prototyping across diverse numerical strategies From strong-form MWR and SPH to Kansa’s RBF collocation and generalized moving least squares (GMLS) - the 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 𝗰𝗼𝗻𝗰𝗲𝗽𝘁𝘂𝗮𝗹𝗹𝘆 𝗰𝗮𝗽𝘁𝘂𝗿𝗲𝘀 𝗮 𝘄𝗶𝗱𝗲 𝘀𝗽𝗲𝗰𝘁𝗿𝘂𝗺 𝗼𝗳 𝘁𝗿𝘂𝗹𝘆 𝗺𝗲𝘀𝗵-𝗳𝗿𝗲𝗲 𝗺𝗲𝘁𝗵𝗼𝗱𝘀. 👉 If you care about 𝘀𝗰𝗮𝗹𝗮𝗯𝗹𝗲 𝗵𝗶𝗴𝗵-𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝘀𝗶𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻𝘀, 𝗺𝗲𝘀𝗵𝗹𝗲𝘀𝘀 𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗹𝗶𝘁𝘆, 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗳𝘂𝘁𝘂𝗿𝗲 𝗼𝗳 𝗶𝗻𝘁𝗲𝗿𝗼𝗽𝗲𝗿𝗮𝗯𝗹𝗲 𝘀𝗼𝗹𝘃𝗲𝗿𝘀, this is one to watch out for! 📝 Presentation slides: https://lnkd.in/eC_eXnn6 𝗘𝗻𝗷𝗼𝘆! #OpenSource #OpenFOAM #CFD #Technology #Meshless #HPC #SimulationEngineering #OpenScience #SPH #RBF #GMLS #NHR4CES #ModularSoftware #StrongFormCFD #ComponentArchitecture #TUDarmstadt #IANUS #IANUSSimulation IANUS Simulation GmbH Technische Universität Darmstadt

277

9 Comments

Today’s guide by Reza Bahmanzadeh, AI Engineer on our Professional Services Team, teaches the essentials of serving large models with vLLM: https://lnkd.in/eHK9nwZm It builds from foundational neural network concepts, like transformers and attention, to introduce practical inference workflows, explore useful vLLM features and offer hands-on, operational guidance for efficient deployments. #LLMs #Serving #Inference #TechGuides

35

1 Comment

Sciloop (YC F25) is an AI Scientist that automates the end-to-end scientific research loop — from generating hypotheses to running experiments and producing results. Founders Bilal Asmatullah and Osman Siddique, IPhO medalists who spent four years researching at MIT CSAIL, repeatedly saw a core bottleneck: even with tools like Cursor and AI copilots, researchers still spend most of their time setting up experiments, monitoring logs, debugging failures, and manually analyzing outputs instead of advancing ideas. Sciloop handles the loop itself. A researcher provides a question or intent, and Sciloop autonomously generates hypotheses, designs experiments, runs them on managed compute in parallel, monitors execution, recovers from failures, and analyzes results. It then recommends next steps and drafts a paper summarizing methods and findings. The goal is simple: turn slow, manual iteration into fast, automated discovery — giving scientists more time to think, explore, and invent. Try it out at app.sciloop.dev. Congrats on the launch, Bilal and Osman! 🚀 https://lnkd.in/gf-C4-KB

952

131 Comments

New paper to be presented at the International Conference on Learning Representations (ICLR): Gradient-Based Diversity Optimization with Differentiable Top-K Objective, by Tianyi Zhou, Sebastian Dalleiger, Ece Calikus & Aristides Gionis. Available here: https://lnkd.in/dmE7DXUH

8

1 Comment

New research presented at NeurIPS 2025 explores how to build fairer and more sustainable data ecosystems for large language models. The study, coauthored by AIMSEC Faculty member Beibei L. of Carnegie Mellon’s Carnegie Mellon University - Heinz College of Information Systems and Public Policy, proposes a framework for LLM data markets that evaluates dataset contributions and introduces a “fairshare” pricing mechanism, highlighting how transparent and equitable pricing could strengthen training data markets while improving model performance and long-term data supply. You can read more here! https://lnkd.in/erSeijPt

1