- Ai - Engineering Mailing Lists

AI Seminar: Daniel Lowd
by Tadepalli, Prasad 13 Mar '25

13 Mar '25

Hi all, Our last AI seminar for this term is on Friday, March 14th. Title: The Manchurian Classifier: Invisible Triggers and Vulnerabilities in Text Models<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…> Speaker: Daniel Lowd, Professor of Computer Science, University of Oregon Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Abstract: Machine learning models are increasingly deployed in critical applications, yet their vulnerability to data poisoning remains a significant security concern. In this talk, I will explore backdoor attacks on text classifiers - a form of adversarial manipulation where hidden "trigger patterns" are embedded during training, causing models to behave normally on standard inputs but produce targeted misclassifications when triggers are present. I will first introduce the concept of backdoor attacks in the text domain and demonstrate why they represent a unique challenge compared to traditional adversarial examples. Then, I will present LLMBkd, our method that leverages large language models to insert backdoor triggers as subtle stylistic variations in text. These attacks are particularly concerning as they require minimal technical expertise while offering attackers a wide range of possible trigger mechanisms that can bypass conventional inspection. Building on this foundation, I will discuss AttrBkd, our more advanced approach that achieves even greater subtlety by extracting fine-grained attributes from existing backdoor attacks. Our comprehensive human evaluations reveal a critical insight: these attacks remain highly effective while being virtually indistinguishable from normal text to human reviewers. This disconnect between human perception and automated detection metrics exposes fundamental limitations in current defensive measures. Finally, I'll discuss the current state of defenses against backdoor attacks and broader security implications for machine learning systems that process text. If time permits, I'll also introduce our ongoing work on Malicious Programming Prompt (MaPP) attacks, which demonstrate how even state-of-the-art coding assistants can be manipulated to produce security vulnerabilities through carefully crafted prompts Biography: Daniel Lowd is a Professor of Computer Science at the University of Oregon. His research interests include machine learning and artificial intelligence, focusing on adversarial methods and statistical relational models. He received his Ph.D. in 2010 from the University of Washington. He has received a Google Faculty Award, an ARO Young Investigator Award, and a best paper award from DEXA. AI Seminars Link: https://engineering.oregonstate.edu/EECS/research/AI-seminars

1 0

AI Seminar: Human-Robot Collaboration for Everyday Household Tasks
by Abou Mrad, Christian 06 Mar '25

06 Mar '25

Dear all, Our next AI seminar is scheduled for Friday, March 7th. Talk details: AI Seminar: Human-Robot Collaboration for Everyday Household Tasks Speaker: Jacob Krantz, Research Scientist, Meta Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: I want a robot assistant that can work with me to accomplish everyday tasks around the home: tidying up the living room, preparing the table for dinner, etc. This broad capability involves the use of natural language for task specification, embodied multi-agent planning, and robust skill execution. Toward this end, I will present our simulation benchmark called PARTNR: Planning And Reasoning Tasks in humaN-Robot collaboration. PARTNR stands as the largest benchmark of its kind, comprising 100,000 natural language tasks spanning 60 houses and 5,819 unique objects. We analyze state-of-the-art LLMs as planners for these tasks, and reveal limitations in coordination, task tracking, and failure recovery. When paired with real human partners, these LLMs are less efficient than human-human collaboration by 1.5x and less efficient than a single human by 1.1x. Through the PARTNR benchmark, many directions of research can be pursued, for example multi-agent planning, 3D scene understanding, skill coordination, Sim2Real transfer, and even HRI studies. Speaker Bio: Jacob Krantz is a research scientist at the Fundamental AI Research lab at Meta (FAIR), where he works on systems of embodied intelligence with applications toward socially intelligent robotics. In the past he co-organized several community challenges relating to semantically driven navigation (VLN, ImageNav). He received his PhD in Computer Science from Oregon State University in 2023 under Dr. Stefan Lee and his BS in Computer Science with a minor in Physics from Gonzaga University in 2019. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Occam Gradient Descent
by Abou Mrad, Christian 27 Feb '25

27 Feb '25

Dear all, Our next AI seminar is scheduled for Friday, February 28th. Talk details: AI Seminar: Occam Gradient Descent Speaker: Balas Natarajan Kausik Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: Gradient descent on large neural networks such as transformers overfits early with many small weights, and therefore inefficient on both computing resources and training data. In response to these inefficiencies, we exploit learning theory to derive Occam Gradient Descent, which interleaves optimal reduction of model size to reduce test loss, with gradient descent on model weights to reduce training loss. Our experiments show that with respect to loss, compute and model size, Occam Gradient Descent outperforms traditional gradient descent across a range of problem domains. Speaker Bio: B.N. “Nat” Kausik is an independent researcher and entrepreneur. A graduate of Cornell, Princeton and IIT Madras, he has worked in both academia and industry. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Towards Intrinsically Explainable Clustering Algorithms
by Abou Mrad, Christian 20 Feb '25

20 Feb '25

Dear all, Our next AI seminar is scheduled for Friday, February 21th. Talk details: AI Seminar: Towards Intrinsically Explainable Clustering Algorithms Speaker: Dr. Eldan Cohen, Assistant Professor of Industrial Engineering at The University of Toronto, The director of the Optimization and Machine Learning (OptiMaL) Lab Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: As machine learning has become more prevalent in society in recent years, the need for trustworthy models that stakeholders can audit has increased dramatically. In particular, one desirable aspect of trustworthiness in ML is that the approaches utilized are constrained so that their predictive mechanisms are innately explainable to humans. Cluster analysis is a fundamental task aims to uncover meaningful groups within data, with many high-stakes applications. Despite the growing interest in explainability in ML, relatively limited work has focused on explainable clustering and most clustering algorithms do not provide any explanation of the obtained partition, requiring post-hoc analysis to characterize the clusters. In this talk, I will describe recent work on clustering algorithms that construct explainable mappings from samples to clusters. We will discuss deep neural approaches, as well as approaches based on combinatorial optimization, to construct such mappings. We will also consider the ability to use clustering constraints to incorporate domain knowledge and improve clustering accuracy and present insight into the trade-off between explainability and the satisfaction of such constraints. Speaker Bio: Eldan Cohen is an Assistant Professor of Industrial Engineering at The University of Toronto and the director of the Optimization and Machine Learning (OptiMaL) Lab. His research interests include machine and deep learning, heuristic search and optimization, and natural language processing, with an emphasis on human compatible approaches. In addition, he has worked on applications of these techniques in various domains such as healthcare, manufacturing, and autonomous agents. He completed his PhD at the Department of Mechanical and Industrial Engineering at the University of Toronto and was a postdoctoral fellow at the department of Computer Science and the Vector Institute for Artificial Intelligence. All information can be found: https://events.oregonstate.edu/event/towards-intrinsically-explainable-clus…<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fevents.or…> For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Understanding the promises and limits of fine-tuning
by Abou Mrad, Christian 13 Feb '25

13 Feb '25

Dear all, Our next AI seminar is scheduled for Friday, February 14th. Talk details: AI Seminar: Understanding the promises and limits of fine-tuning Speaker: Dr. Aditi Raghunathan, Assistant Professor of Computer Science, Carnegie Mellon University Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: In recent years, foundation models—large pretrained models that can be adapted for a wide range of tasks—have achieved state-of-the-art performance on a variety of tasks. The adaptation or fine-tuning process is a crucial component that enables specialization to the task of interest, and is the de facto standard for mitigating risks such as reducing toxic and harmful generations from large language models. While the pretrained models are trained on broad data, the adaptation (or fine-tuning) process is often performed on limited well-curated data. How well does fine-tuning generalize beyond the narrow training distribution? Via theory and experiments, we show how to improve current fine-tuning approaches so that they can better leverage diverse pretraining knowledge and improve downstream performance across broader settings than the narrow fine-tuning data. On the flip side, we show that pretrained knowledge can be hard to get rid of, thereby underlining the potential perils of over reliance on fine-tuning for safety. Speaker Bio: Aditi Raghunathan is an Assistant Professor in the Computer Science Department at CMU. She received her PhD from Stanford in 2021, and Bachelor of Technology from IIT Madras in 2016. She is a recipient of the Okawa research grant, Schmidt AI2050 Early Career Fellowship, the Google Research Scholar Award, Rising Stars in EECS, Google PhD Fellowship, Open Philanthropy AI Fellowship, Stanford School of Engineering Fellowship and Google Anita Borg Memorial Fellowship. She was featured in the Forbes 30 under 30 list for her contributions to reliable machine learning. Her PhD thesis was awarded the Arthur Samuel Best Thesis Award at Stanford. Her research has also been recognized by multiple orals and spotlights at top conferences, and a Best Paper Award at Data Problems in ML Workshop at ICLR 2024. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Active Learning For Human Preference Elicitation In LLMs
by Abou Mrad, Christian 06 Feb '25

06 Feb '25

Dear all, Our next AI seminar is scheduled for Friday, February 7th. Talk details: AI Seminar: Active Learning For Human Preference Elicitation In LLMs Speaker: Branislav Kveton, Adobe Research Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: Learning from human preferences has been central to recent advances in AI. Motivated by the cost of obtaining high-quality human feedback, we study active learning of large language models (LLMs) through adaptive human preference elicitation. The key idea in our work is to linearize LLMs, either the reward model or the optimized policy, and apply optimal designs, a methodology for computing optimal information-gathering policies in statistics. We design efficient algorithms, fundamentally leverage the autoregressive nature of the LLMs, and evaluate our algorithms on state-of-the-art models. Speaker Bio: Branislav (Brano) is a Principal Research Scientist at Adobe Research and leads their team on planning and reasoning with generative AI models. He proposes, analyzes, and applies algorithms that learn incrementally, run in real time, and converge to near-optimal solutions with more feedback. Brano has been an active researcher in the fields of ML and AI for more than 15 years. His works have been published at top-tier conferences in these fields and he serves on senior roles at these conferences. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Introduction To Accelerated Computing with NVDIA GPUs
by Abou Mrad, Christian 31 Jan '25

31 Jan '25

Dear all, Two AI seminars are scheduled for Friday, January 31th. This is the second talk. Talk details: AI Seminar: Introduction To Accelerated Computing with NVDIA GPUs Speaker: Zoe Ryan, Solutions Architect, NVIDIA Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: In this session we will cover the basics of the NVIDIA GPU (graphics processing unit) and how it can be used in the context of education and research. We will start with an overview of how the GPU functions compared to traditional CPU computing and will follow that with an outline of the existing hardware portfolio. Then we will transition to talk about how you can program for the GPU, and what workloads are well suited to take advantage of the parallel nature of the GPU. Finally, we will have a hands-on demonstration of cutting edge LLMs (large language models) and generative AI workloads that rely on the compute power of the GPU. Speaker Bio: Zoe Ryan is a solutions architect at NVIDIA. She's worked in embedded systems/simulations and GPU-accelerated high-performance computing, specifically targeting higher education institutions and researchers to adopt and fully utilize GPU technology. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Painful Intelligence: What AI Can Tell Us About Human Suffering
by Abou Mrad, Christian 31 Jan '25

31 Jan '25

Dear all, Two AI seminars are scheduled for Friday, January 31th. This is the first talk. Talk details: AI Seminar: Painful Intelligence: What AI Can Tell Us About Human Suffering Speaker: Dr. Aapo Hyvärinen, Professor of Computer Science, University of Helsinki Time: 10:00 AM Location: Zoom remotely only Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: This talk introduces my recent e-book with the same title, freely available on Arxiv<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.cs.he…>. The book uses the modern theory of artificial intelligence (AI) to understand human suffering or mental pain. Both humans and sophisticated AI agents process information about the world in order to achieve goals and obtain rewards, which is why AI can be used as a model of the human brain and mind. The book starts with the assumption that suffering is mainly caused by frustration. Frustration means the failure of an agent (whether AI or human) to achieve a goal or a reward it wanted or expected. Frustration is inevitable because of the overwhelming complexity of the world, limited computational resources, and scarcity of good data. In particular, such limitations imply that an agent acting in the real world must cope with uncontrollability, unpredictability, and uncertainty, which all lead to frustration. Such computational theory is finally used to derive various interventions or training methods that will reduce suffering in humans. The ensuing interventions are very similar to those proposed by Buddhist and Stoic philosophy, and include mindfulness meditation. Speaker Bio: Aapo Hyvärinen studied undergraduate mathematics at the universities of Helsinki (Finland), Vienna (Austria), and Paris (France), and obtained a Ph.D. degree in Information Science at the Helsinki University of Technology in 1997. After post-doctoral work at the Helsinki University of Technology, he moved to the University of Helsinki in 2003, where he was appointed Professor in 2008, at the Department of Computer Science. From 2016 to 2019, he was Professor of Machine Learning at the Gatsby Computational Neuroscience Unit, University College London, UK. Aapo Hyvarinen is the main author of the books "Independent Component Analysis" (2001), "Natural Image Statistics" (2009), and "Painful Intelligence" (2022). He is Action Editor at the Journal of Machine Learning Research and Neural Computation, and has worked as Area Chair at ICML, ICLR, AISTATS, UAI, ACML and NeurIPS. He is a Fellow at the European Laboratory for Learning and Intelligent Systems (ELLIS). His research focuses on the probabilistic theory of machine learning and its applications in neuroscience. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1

AI Seminar: Painful Intelligence: What AI Can Tell Us About Human Suffering
by Abou Mrad, Christian 28 Jan '25

28 Jan '25

Dear all, Two AI seminars are scheduled for Friday, January 31th. This is the first talk. Talk details: AI Seminar: Painful Intelligence: What AI Can Tell Us About Human Suffering Speaker: Dr. Aapo Hyvärinen, Professor of Computer Science, University of Helsinki Time: 10:00 AM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: This talk introduces my recent e-book with the same title, freely available on Arxiv<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.cs.he…>. The book uses the modern theory of artificial intelligence (AI) to understand human suffering or mental pain. Both humans and sophisticated AI agents process information about the world in order to achieve goals and obtain rewards, which is why AI can be used as a model of the human brain and mind. The book starts with the assumption that suffering is mainly caused by frustration. Frustration means the failure of an agent (whether AI or human) to achieve a goal or a reward it wanted or expected. Frustration is inevitable because of the overwhelming complexity of the world, limited computational resources, and scarcity of good data. In particular, such limitations imply that an agent acting in the real world must cope with uncontrollability, unpredictability, and uncertainty, which all lead to frustration. Such computational theory is finally used to derive various interventions or training methods that will reduce suffering in humans. The ensuing interventions are very similar to those proposed by Buddhist and Stoic philosophy, and include mindfulness meditation. Speaker Bio: Aapo Hyvärinen studied undergraduate mathematics at the universities of Helsinki (Finland), Vienna (Austria), and Paris (France), and obtained a Ph.D. degree in Information Science at the Helsinki University of Technology in 1997. After post-doctoral work at the Helsinki University of Technology, he moved to the University of Helsinki in 2003, where he was appointed Professor in 2008, at the Department of Computer Science. From 2016 to 2019, he was Professor of Machine Learning at the Gatsby Computational Neuroscience Unit, University College London, UK. Aapo Hyvarinen is the main author of the books "Independent Component Analysis" (2001), "Natural Image Statistics" (2009), and "Painful Intelligence" (2022). He is Action Editor at the Journal of Machine Learning Research and Neural Computation, and has worked as Area Chair at ICML, ICLR, AISTATS, UAI, ACML and NeurIPS. He is a Fellow at the European Laboratory for Learning and Intelligent Systems (ELLIS). His research focuses on the probabilistic theory of machine learning and its applications in neuroscience. If you would like to meet with the speakers one-on-one, please email tadepall(a)oregonstate.edu<mailto:tadepall@oregonstate.edu> by Tuesday. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 0

AI Seminar: "Explainable AI with Guarantees" by Dr. Adnan Darwiche
by Abou Mrad, Christian 23 Jan '25

23 Jan '25

Dear all, Our next AI seminar is scheduled for Friday, January 24th. Talk details: AI Seminar: Explainable AI with Guarantees Speaker: Dr. Adnan Darwiche, Professor of Computer Science, UCLA Time: 2:00 PM Location: KEC 1001 and Zoom Zoom link: https://oregonstate.zoom.us/s/98357211915<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonsta…> Talk Abstract: I will discuss a theory of formal explainable AI that has emerged over the last few years, which is geared towards explaining the decisions of learned classifiers likes ones based on Bayesian Networks, random forests and some types of neural networks. The theory is based on compiling the input-output behavior of such classifiers into symbolic form and then using sophisticated machinery from symbolic logic to explain decisions. The theory employs newly-introduced, and profound, abstraction operators that distill the essence of an instance (i.e., classifier input) which causes the decision on that instance. Among the implications is an ability to identify minimal aspects of an instance that guarantee the decision, and minimal aspects that must be changed (and in what way) to change the decision. I will illustrate the theory using concrete examples and case studies and also tie these developments to more classical explanations and some widely used approximate explanations. Speaker Bio: Adnan Darwiche is a professor and former chairman of the computer science department at UCLA. He directs the Automated Reasoning Group, which focuses on symbolic reasoning, probabilistic reasoning and their applications to machine learning. Professor Darwiche is a Fellow of AAAI and ACM and recipient of the Lockheed Martin Excellence in Teaching Award. He is a former editor-in-chief of the Journal of Artificial Intelligence Research (JAIR) and author of “Modeling and Reasoning with Bayesian Networks,” by Cambridge University Press. Many of his works can be found on his YouTube Channel, https://www.youtube.com/@UCLA.Reasoning<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.youtu…>. If you would like to meet with the speakers one-on-one, please email tadepall(a)oregonstate.edu<mailto:tadepall@oregonstate.edu>. For future AI seminars, please visit: https://engineering.oregonstate.edu/EECS/research/AI-seminars<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fengineeri…>. Best, Christian Abou Mrad Graduate Student Oregon State University

1 1