Aidan Gomez

NiNo: A Novel Machine Learning Approach to Accelerate Neural Network Training through Neuron Interaction and Nowcasting Researchers from Samsung’s SAIT AI Lab, Concordia University, Université de Montréal, and Mila have introduced a novel approach known as Neuron Interaction and Nowcasting (NINO) networks. This method aims to significantly reduce training time by predicting the future state of network parameters. Rather than applying an optimization step at every iteration, as with traditional methods, NINO employs a learnable function to predict future parameter updates periodically. By integrating neural graphs—which capture the relationships and interactions between neurons within layers—NINO can make rare yet highly accurate predictions. This periodic approach reduces the computational load while maintaining accuracy, particularly in complex architectures like transformers. At the core of the NINO methodology lies its ability to leverage neuron connectivity through graph neural networks (GNNs). Traditional optimizers like Adam treat parameter updates independently without considering the interactions between neurons. NINO, however, uses neural graphs to model these interactions, making predictions about future network parameters in a way that reflects the network’s inherent structure. The researchers built on the Weight Nowcaster Networks (WNN) method but improved it by incorporating neuron interaction modeling. They conditioned NINO to predict parameter changes for the near and distant future. This adaptability allows NINO to be applied at different stages of training without requiring constant retraining, making it suitable for various neural architectures, including vision and language tasks. The model can efficiently learn how network parameters evolve by using supervised learning from training trajectories across multiple tasks, enabling faster convergence.... Read our article on this:https://lnkd.in/gSQFYtcG Paper: https://lnkd.in/g9rJ5JT2 GitHub Page: https://lnkd.in/gJPbT-kM

25

One of my favourite podcasts in the past few months. Peter Attia is both a medical doctor and and engineer. In this podcast, he discusses AI with Isaac Kohane, a medical doctor with a PhD in Computer Science. The first half they do a great job of exploring the history of AI and how we got today. The second half dives into AI in medicine. There is huge potential to boost medical care with further application of AI that exists today. https://lnkd.in/gZK3tcHc

6

I'm excited that this week at the CAIAC - Canadian Artificial Intelligence Association conference George Wang will be presenting a paper Adaptive Learning Rates for Gradient Boosting Trees, based on work he did as an intern at Kinaxis last fall, in collaboration with Christopher Wang, Yunfei O., Behrouz Soleimani, PhD. There are 3 reasons this is both impressive and cool: 1. They combine heuristics, machine learning, and optimization to speed up the training for an important algorithm, Gradient Boosted Machines, widely used in both academia and industry. Faster training improves model performance. 2. This paper is just one of the many sparks of innovation in #artificialintelligence coming out of Kinaxis, where > 50% of our patents filed in the last several years have been in this space, because we continue to push the boundaries by inventing new solutions like this one to solve problems our customers face. We've been delivering AI to the market since 2018 and were the first to provide solutions powered by AI to address both demand and supply. We are rooted in Canada, which brought the world out of the last AI winter in 2012 when a team from the University of Toronto won the famous Imagenet competition by blowing the lid off previous results. Their submission was layers of neural networks called deep learning, which later became the foundation for the #generativeAI revolution we are seeing today. 3. We have an outstanding intern program in #machinelearning at Kinaxis, where interns work on real problems, guided by senior staff. As a testament to letting the students shine, when their paper was accepted they decided to have George present it. So check out the paper if you're at the conference, but if you miss the event you can't miss the fact that we are actively speaking and sparking on AI. In fact, just last week Chantal Bisson-Krol spoke in Edmonton at the #upperbound conference on AI-Powered Strategies for Resilient Supply Chains and I spoke at the Reuters Supply Chain USA event on Embedding AI in supply chain to deliver end-to-end value.

43

2 Comments

CppNorth, The Canadian C++ Conference 2024 • July 21-24, 2024 • Toronto Preview: Conor Hoekstra: "Composition Intuition II" Please Like, Follow, & Share: CppNorth, The Canadian C++ Conference  https://lnkd.in/g6x2ktpP This sequel to Composition Intuition will be a masterclass about everything you wanted to know (and potentially what you didn't even know you wanted to know) when it comes to function composition and combinators. We will cover the different "composition strategies" used in languages like C++, Python, Haskell, APL, J, KAP, Jelly, Uiua, Clojure and more. Furthermore, we will in more detail cover the most common combinators and the patterns in which they are most useful. Registration is open: store.cppnorth.ca (select 'Early Bird') Make your room reservations now: https://lnkd.in/g-2dePpQ

19

🔬 Researchers from Lawrence Berkeley National Laboratory (Berkeley Lab), UC Berkeley, the Institute for Systems Biology, and Université Laval have unlocked new insights into the TIP60 protein complex, a critical gateway to DNA accessibility. Published in the journal Science, this groundbreaking study reveals the atomic structure of TIP60, paving the way for potential breakthroughs in targeted therapies for diseases such as cancer and Alzheimer's. Using advanced high-resolution cryo-electron microscopy, the team elucidated how this complex enables essential DNA repair and transcription processes, which are vital for cellular health and often disrupted in disease states. This research not only enhances our understanding of genetic regulation but also opens exciting avenues for developing therapies that manipulate TIP60's activity to treat genetic and neurodegenerative disorders. Explore how this significant discovery could transform future medical treatments! #Genetics #HealthcareInnovation #ScienceNews Read more: https://lnkd.in/eMnjr2bp

3

There's only 5 days left to submit your paper! Remember, we only a short 1-page abstract

2

💥 HSLU at NeurIPS? Oh yes we are!! Congrats to Fabian Gröger and team for being accepted to one of the highest impact AI/ML conferences on the planet. 🚀 🎉 🌎 NeurIPS 2024, the Thirty-Eighth Annual Conference on Neural Information Processing Systems, will be held at the Vancouver Convention Center Monday Dec 9 through Sunday Dec 15. HSLU researchers Fabian Gröger, Simone Lionetti, Ludovic Amruthalingam and Marc Pouly together with their collaborators from University of Basel, University Hospital Basel and Northwestern University are presenting “SelfClean: A Self-Supervised Data Cleaning Strategy”. ⚡ PROBLEM Most benchmark datasets for computer vision contain irrelevant images, near duplicates, and label errors. Consequently, model performance on these benchmarks may not be an accurate estimate of generalization capabilities. This is a particularly acute concern in computer vision for medicine where datasets are typically small, stakes are high, and annotation processes are expensive and error-prone. 😎 SOLUTION The researchers have developed a general procedure to clean up image datasets exploiting a latent space learned with self-supervision. By relying on self-supervised learning, the approach focuses on intrinsic properties of the data and avoids annotation biases. SelfClean achieves state-of-the-art performance in detecting irrelevant images, near duplicates, and label errors within popular computer vision benchmarks. 🌐 You can read the pre-print here: https://lnkd.in/eAQidcj4 👩‍💻 The official implementation is here: https://lnkd.in/eanJnQ73 🌟 🤩 Amazing work guys! Congratulations, and enjoy the trip to Canada. Philippe Gottfrois, Alvaro Gonzalez Jimenez, Matthew Groh, Alexander Navarini, LAC² - Lucerne AI and Cognitive Community, Switzerland Global Enterprise, Switzerland Innovation Park Central, Lucerne Business, Economic Promotion Canton of Zug, Schlössli Collab, Lucerne University of Applied Sciences and Arts, HSLU – Hochschule Luzern – Informatik

50

1 Comment

🚨 Call for Papers: 3rd International Workshop on Responsible AI Engineering (#RAIE2025) 🚨 📝 Submit your paper by November 11, 2024 and join us in shaping the future of responsible AI at RAIE 2025, part of the #ICSE2025 conference in Ottawa, Canada! 👉 https://lnkd.in/edHqtA29 This workshop will explore the challenges of building AI systems that are not only innovative but also trustworthy, ethical, and safe in their real-world applications. As AI technologies, especially large language models (LLMs), continue to evolve, the need for responsible AI has never been more pressing. 👨‍💻 Dr. Maximilian Poretschkin, Head of the Trustworthy AI Innovation Group at the Lamarr Institute and Head of AI assurance and certification at Fraunhofer IAIS, is part of the organizing committee and will help lead discussions on developing transparent and trustworthy AI systems that align with human values. 💡 Whether you're focused on AI governance, safety regulations, or new verification tools, this workshop offers the perfect platform to share and gain insights. ℹ️ Learn more about the workshop here: https://lnkd.in/eEd7_mTh  #cfp #responsibleai #aisafety #aiengineering #softwareengineering | Qinghua Lu (CSIRO's Data61) Foutse Khomh (Polytechnique Montréal) Apostol Vassilev (National Institute of Standards and Technology (NIST)) Muneera Bano (CSIRO's Data61)

30

2 Comments

This is so exciting. For the Nobel body, for the scientific community, and for the world to acknowledge the tremendous work done to enable AI and ML as we know it today. The shift in paradigm that was created with Jeffrey's work (and the co-recipient's) has made work in this space, deep, philosophical and applicable at same time. https://lnkd.in/ggW-q45K #machinelearning #ai #artificialintelligence Congratulations form all of us at Mesh AI (collaborative scheduling to end clinician burnout).

36

2 Comments

Dr. Tali Dekel, a leading researcher in generative AI from the Weizmann Institute of Science, is pushing the boundaries of what's possible with large-scale deep-learning models! Her collaboration with Google has led to the creation of Lumiere, a groundbreaking text-to-video model for video generation. 🌟 Lumiere can transform simple text prompts or reference photos into rich, realistic videos. Imagine a squirrel on the Moon or a yawning Mona Lisa – Lumiere makes it happen! Dr. Dekel's work focuses on "re-rendering reality," recreating the visual world using computational tools. Her team's research is revealing fascinating insights and pushing the frontiers of technology! https://lnkd.in/eYKQYzKm #AI #WomenforScience

11

👉🏼 ChatENT: Augmented Large Language Model for Expert Knowledge Retrieval in Otolaryngology-Head and Neck Surgery 🤓 Cai Long 👇🏻 https://lnkd.in/e7_3F-hi 🔍 Focus on data insights: - ChatENT outperformed ChatGPT4.0 in both the Canadian Royal College OHNS sample examination questions challenge and the US board practice questions challenge. - A significant error reduction of 58.4% and 26.0% was observed, respectively. - ChatENT demonstrated enhanced performance in analyzing and interpreting OHNS information, generating fewer hallucinations and showing greater consistency. 💡 Main outcomes and implications: - ChatENT is the first specialty-specific knowledge retrieval artificial intelligence in the medical field utilizing the latest LLM. - Promising applications in medical education, patient education, and clinical decision support are highlighted. - Overcoming limitations of existing LLMs, ChatENT signals a future of more precise, safe, and user-friendly AI applications in OHNS and other medical fields. 📚 Field significance: - Artificial Intelligence - Medical Education - Clinical Decision Support 🗄️: [#Otolaryngology #ArtificialIntelligence #MedicalEducation #ClinicalDecisionSupport]

1

Onur Mutlu will give an invited lecture on Tuesday, May 7 11:00AM EDT on "Memory-Centric Computing" as part of the Department of Computer Science, University of Toronto Distinguished Lecture Series. This lecture is open to the public. No registration is required, but space is limited!   Talk details: https://uoft.me/asG In this talk, we discuss the design and adoption of more intelligent architectures that aim to fundamentally reduce memory latency and energy, and practically enable computation close to data, with at least two promising directions: 1) processing using memory, which exploits analog operational properties of memory chips to perform massively-parallel operations in memory, with low-cost changes, 2) processing near memory, which integrates sophisticated additional processing capability in memory controllers, the logic layer of 3D-stacked memory technologies, or memory chips to enable high memory bandwidth and low memory latency to near-memory logic. We show how these architectures lead to significant improvements in performance and energy consumption of many important workloads, such as graph analytics, database systems, machine learning, video processing, climate modeling, genome analysis. Finally, we discuss how to enable adoption of such intelligent architectures, with guiding principles for future computing architecture and system designs. ETH Zurich, Department of Information Technology and Electrical Engineering #computing #hardware #software #design #paradigms #innovation #research #computerarchitecture #processinginmemory

50

🚀 Looking to streamline fine-tuning for models like Llama 3, Mistral, and Mixtral? Our latest blog shows how Anyscale simplifies LLMs fine-tuning with Ray: 🔑 Key takeaways: • Scalable fine-tuning with LLM-Forge • Model Deployment with RayLLM Read more: https://lnkd.in/gQwD4A2U

34

🚨 Can AI System Quantify Misinformation in Real-Time? Groundbreaking Research Says Yes! 🚨 📊 Researchers at the University of Waterloo have developed a revolutionary system called U-MAS (UbiLab Misinformation Analysis System) to detect, analyze, and quantify health misinformation on social media as it happens. 🔑 Key features of U-MAS: • 🤖 Uses AI and machine learning to analyze Twitter data • 🧠 Performs topic modeling, sentiment analysis, and misinformation classification • 📈 Provides interactive dashboards with real-time misinformation percentages • ⏳ Enables monitoring of misinformation trends and volumes over time 🧪 The system was tested on fluoride-related misinformation, accurately quantifying the percentage of false information in the dataset. It's now being adapted for vaccine hesitancy and heat illness topics. 💡 U-MAS empowers public health officials and researchers to: • Measure the exact prevalence of misinformation on specific health topics • Identify emerging misinformation trends as a percentage of overall conversation • Develop data-driven, targeted interventions based on quantifiable metrics 🌟 This groundbreaking work puts a number on the spread of health misinformation online, potentially revolutionizing how we combat fake news. Read the full paper here: [Link to paper] #HealthMisinformation #AI #PublicHealth #DataAnalytics​​​​​​​​​​​​​​​​ #HealthMisinformation #AI #PublicHealth #DataAnalytics

1

Vector Institute's Researcher Wenhu Chen at University of Waterloo is pushing the boundaries of AI by improving and benchmarking foundation models. These efforts are crucial in developing more efficient, accurate, and scalable models that power a wide range of applications, from language understanding to data processing. Learn more about how Wenhu Chen’s team is shaping the future of AI research. https://lnkd.in/gr8CVrvg #AI #Research #FoundationModels #Innovation #MachineLearning

1

AI is driving powerful possibilities in scientific discovery. Yoshua Bengio, an AI visionary, Turing Award-winner, computer science professor at the Universite de Montreal, and one of our scientific advisors, sat down with Berton Earnshaw and Ben Mabey to discuss the role of AI in drug discovery. 🔹 The takeaway: AI will provide transformative brain power to human scientists – and that’s critical for accelerating our understanding of disease and new therapeutics. ◾ On AI outperforming human scientists: Yoshua says: “I have a very strong belief that we are on the trajectory to build the tools to do as well as human scientists or better. Human brains can’t absorb a billion data points, especially something that is not naturally intuitive to us like millions of images that are not in our day to day, like cells. Although we can generate scientific hypotheses, it’s hard for individual humans to create a systematic map of the different possible interpretations. The whole scientific community in a way is doing something like this, but it’s very difficult for individual humans alone. ◾ On using AI to decide which paths to pursue without bias: Yoshua says: “That whole uncertainty – many possible hypotheses – that is what should drive efficient experimentation. Computers can do the scientific work in a much more systematic and large-scale way than single brains or teams of researchers could do…There’s a good chance within the order of a decade we’ll figure out how to do these things with AI and it’s going to completely transform society and scientific research. Being at the forefront of such a high level of transformation will be highly valuable.” ◾ On using AI to arrive at answers beyond the known science: Berton says: “One of the primary reasons we went after images of cells, and using modern machine learning to create descriptions of morphology that go beyond our ability to handcraft it, is because we expect there to be a lot of information that we do not have the language to properly talk about but that nonetheless gives us insights and allow us to generate therapeutic hypotheses about the biology that we’re observing.” 👉 Watch the extended conversation here: https://lnkd.in/eahedX3f #ai #ml #techbio #drugdiscovery #science #tech #scientists #research

103

🎉 🎉 Check out this #HighlyCitedPaper "An Overview of Nanotechnologies for Drug Delivery to the Brain" by Shawn Wettig et al. author affiliations: University of Waterloo You can find it at: https://shorturl.at/ilrX5

3

Last Friday we hosted an enlightening webinar on "Understanding and Navigating Bias in Machine Learning Models" presented by Dr. Afaf Taïk, a postdoctoral researcher at Mila and Université de Montréal, specializing in responsible AI with a focus on fairness and privacy. During this webinar, Dr. Afaf Taïk delved into the categories of harms caused by unfair machine learning models, exploring the reasons behind these unfairness, which can stem from biased models or biased data. Additionally, Dr. Taïk examined various approaches to defining fairness in machine learning, emphasizing the importance of considering the consequences and context of the model's use case. Finally, Dr. Taïk discussed several techniques to avoid bias and achieve fairness, presenting the pros and cons of each strategy. 🎥 If you missed the webinar or would like to watch it again, please check the link to the recording in the comment section below. We extend our heartfelt thanks to Dr. Afaf Taïk for sharing her expertise and to all attendees for the insightful discussions at the end of the webinar. #MoroccoAI #Webinar #ArtificialIntelligence  #BiasInML #AIFairness

35

1 Comment