- Ai - Engineering Mailing Lists

[CoRIS-List] Updated Schedule: Oregon State Robotics Spring 2022 Seminar Series
by Ross L. Hatton 26 Apr '22

26 Apr '22

Hi All, I am very pleased to announce that the OSU Robotics seminar on May 20 will be presented by Reid Simmons of the Carnegie Mellon University Robotics Institute. (Chris Sanchez from OSU Psychology, has graciously agreed to shift his OSU Robotics seminar to next year.) The full updated schedule for the current term is below. --- 4/1/2022 Cristina Wilson -UPenn GRASP Lab & RAD Lab -Collaborative human-robot teaming -https://www.radlab.us 4/8/2022 RGSA Round Table 4/15/2022 Chris Neider -Starship Technologies -Food delivery robots -https://www.starship.xyz/ 4/22/2022 Kim Ingraham -University of Washington Mechanical Engineering, Rehabilitation Medicine -Biomechanics and wearable robots -https://kim-ingraham.com 4/29/2022 ICRA practice talks 5/6/2022 Shai Revzen -University of Michigan Electrical Engineering -Dynamics, bioinspired, legged locomotion -https://robotics.umich.edu/profile/shai-revzen/ 5/13/2022 No Seminar: NW Robotics Symposium at University of Washington 5/20/2022 Reid Simmons -Carnegie Mellon University Robotics Institute -Robust autonomy -Human-Robot interaction 5/27/2022 No seminar: ICRA 6/3/2022 Edward Wang -UCSD ECE & Design Lab -Design and health monitoring -https://www.ejaywang.com (Kristen Macuga from OSU Psychology was originally scheduled to speak this term, but has rescheduled for the Fall) =========== Ross L. Hatton Associate Professor, Robotics and Mechanical Engineering Collaborative Robotics and Intelligent Systems Institute Oregon State University coris.oregonstate.edu research.engr.oregonstate.edu/lram/ rosslhatton.com ross.hatton(a)oregonstate.edu

1 0

AI seminar: April 27, 2022
by Mangannavar, Rajesh Devaraddi 26 Apr '22

26 Apr '22

Dear all, Our next AI seminar on *"Building neural networks that know what they don’t know" *by Balaji Lakshminarayanan is scheduled to be on April 27th(Tomorrow), 1-2 PM PST (Add to google calendar <https://calendar.google.com/event?action=TEMPLATE&tmeid=YXExYTZuY2NiZXRhbHJ…> ). It will be followed by a 30-minute Q&A session by the graduate students. Zoom Link: https://oregonstate.zoom.us/j/93591935144?pwd=YjZaSjBYS0NmNUtjQzBEdzhPeDZ5U… *Building neural networks that know what they don’t know* Balaji Lakshminarayanan Staff Research Scientist Google Brain *Abstract:* Deep neural networks can make overconfident errors and assign high confidence predictions to inputs far away from the training data. Well-calibrated predictive uncertainty estimates are important to know when to trust a model's predictions, especially for safe deployment of models in applications where the train and test distributions can be different. I'll first present some concrete examples that motivate the need for uncertainty and out-of-distribution (OOD) robustness in deep learning. Next, I'll present an overview of our recent work focused on building neural networks that know what they don’t know: this includes methods which improve single model uncertainty (e.g. spectral-normalized neural Gaussian processes), methods which average over multiple neural network predictions such as Bayesian neural nets and deep ensembles, and methods that leverage better representations (e.g. improving “near-OOD” detection). *Speaker Bio:* Balaji Lakshminarayanan is a staff research scientist at Google Brain. His recent research is focused on probabilistic deep learning, specifically, uncertainty estimation, out-of-distribution robustness and applications. Before joining Google Brain, he was a research scientist at DeepMind. He received his Ph.D. from the Gatsby Unit, University College London and Master’s degree from Oregon State University. He has co-organized several workshops on "Uncertainty and Robustness in deep learning" and served as Area Chair for NeurIPS, ICML, ICLR, and AISTATS. This talk will be followed by a talk on the topic "*Probabilistic Planning through the Lens of Approximate Inference*" by professor Roni Khardon on 4th May (Wednesday) 1-2 PM, 2022 *Please watch this space for future AI Seminars :* * https://eecs.oregonstate.edu/ai-events <https://eecs.oregonstate.edu/ai-events>* Rajesh Mangannavar, Graduate Student Oregon State University ---- AI Seminar Important Reminders: -> For graduate students in the AI program, attendance is strongly encouraged.

1 0

[CoRIS-List] Reminder: OSU Robotics Seminar Today: Kim Ingraham, "Enabling human mobility through personalized robotic control"
by Ross L. Hatton 22 Apr '22

22 Apr '22

Hi All, As a reminder, this week's OSU Robotics seminar will be presented by Kim Ingraham from the University of Washington, on "Enabling human mobility through personalized robotic control". The seminar will be from 10-11am, followed by an 11-11:30 student-only Q&A session. The seminar will be an *in person* event, simulcast over Zoom, and will be immediately followed by a half-hour students-only Q&A session with the speaker. April 22, 10am LINC 268 https://oregonstate.zoom.us/j/93908195212?pwd=cEdoQVdac0JxMUpOSS9xVzZjVG5xU… <https://oregonstate.zoom.us/j/93908195212?pwd=cEdoQVdac0JxMUpOSS9xVzZjVG5xU…> ----------------- Seminar Title: Enabling human mobility through personalized robotic control Abstract: Assistive robotic technologies—like protheses, exoskeletons, and semi-autonomous powered wheelchairs—have the potential to meaningfully transform human mobility. Yet, despite great strides in the development of wearable assistive robots over the last several decades, people are not widely using these systems in their daily lives. Fundamentally, we do not yet know how to apply robotic assistance to the human body in order to promote meaningful clinical improvements or achieve targeted physiological goals. To this end, my work is focused on the design and experimental evaluation of personalized, adaptive control strategies for assistive robotic devices. In this seminar, I will present work that advances our understanding of how to provide robotic assistance to users outside the laboratory environment. As an example of a personalized control system, I will demonstrate that exoskeleton users can quickly and precisely identify features of robotic assistance that they prefer, and highlight characteristics of user preference that make the design of personalized control systems both compelling and challenging. While many assistive robots support users during walking, I will also share insights into how these methods may be applied to wheeled mobility technologies, specifically for children with disabilities. Together, this work supports my future research goal of designing personalized, adaptive control strategies for wearable assistive robots in order to enable people to meet their goals and achieve full participation in their daily lives. Bio: Kim Ingraham (she/her) is a post-doctoral fellow at the University of Washington in the departments of Rehabilitation Medicine and Mechanical Engineering, and is a member of the multidisciplinary group CREATE (Center for Research and Education on Accessible Technology and Experiences). Kim holds degrees in Mechanical Engineering from the University of Michigan (PhD, 2021) and in Biomedical Engineering from Vanderbilt University (BE, 2012). Prior to beginning graduate school, she worked on an interdisciplinary team of scientists, engineers, and clinicians as a Research Engineer at the Shirley Ryan AbilityLab (2012-2015). Kim is passionate about using assistive wearable robots to advance human mobility in all forms. In her research, she has developed and evaluated physiologically-inspired control systems for a variety of assistive devices, including powered lower-limb prostheses, robotic exoskeletons, and powered wheelchairs for young children with disabilities. Kim is an NSF graduate research fellow, and served as both a graduate student instructor and an engineering teaching consultant at the University of Michigan. =========== Ross L. Hatton Associate Professor, Robotics and Mechanical Engineering Collaborative Robotics and Intelligent Systems Institute Oregon State University coris.oregonstate.edu <http://coris.oregonstate.edu/> research.engr.oregonstate.edu/lram/ rosslhatton.com ross.hatton(a)oregonstate.edu

1 0

AI seminar: April 27, 2022
by Mangannavar, Rajesh Devaraddi 21 Apr '22

21 Apr '22

Dear all, Our next AI seminar on *"Building neural networks that know what they don’t know" *by Balaji Lakshminarayanan is scheduled to be on April 27th, 1-2 PM PST. It will be followed by a 30-minute Q&A session by the graduate students. Zoom Link: https://oregonstate.zoom.us/j/93591935144?pwd=YjZaSjBYS0NmNUtjQzBEdzhPeDZ5U… *Building neural networks that know what they don’t know* Balaji Lakshminarayanan Staff Research Scientist Google Brain *Abstract:* Deep neural networks can make overconfident errors and assign high confidence predictions to inputs far away from the training data. Well-calibrated predictive uncertainty estimates are important to know when to trust a model's predictions, especially for safe deployment of models in applications where the train and test distributions can be different. I'll first present some concrete examples that motivate the need for uncertainty and out-of-distribution (OOD) robustness in deep learning. Next, I'll present an overview of our recent work focused on building neural networks that know what they don’t know: this includes methods which improve single model uncertainty (e.g. spectral-normalized neural Gaussian processes), methods which average over multiple neural network predictions such as Bayesian neural nets and deep ensembles, and methods that leverage better representations (e.g. improving “near-OOD” detection). *Speaker Bio:* Balaji Lakshminarayanan is a staff research scientist at Google Brain. His recent research is focused on probabilistic deep learning, specifically, uncertainty estimation, out-of-distribution robustness and applications. Before joining Google Brain, he was a research scientist at DeepMind. He received his PhD from the Gatsby Unit, University College London and Master’s degree from Oregon State University. He has co-organized several workshops on "Uncertainty and Robustness in deep learning" and served as Area Chair for NeurIPS, ICML, ICLR and AISTATS. *Please watch this space for future AI Seminars :* * https://eecs.oregonstate.edu/ai-events <https://eecs.oregonstate.edu/ai-events>* Rajesh Mangannavar, Graduate Student Oregon State University ---- AI Seminar Important Reminders: -> For graduate students in the AI program, attendance is strongly encouraged.

1 0

AI Alignment Reading Group
by Pope, Quintin 21 Apr '22

21 Apr '22

Hello everyone, We’ll be holding the next AI alignment reading group meeting at 2 PM PST this Friday. Our paper will be “Active Learning Helps Pretrained Models Learn the Intended Task <https://arxiv.org/abs/2204.08491>”. Models can fail in unpredictable ways during deployment due to task > ambiguity, when multiple behaviors are consistent with the provided > training data. An example is an object classifier trained on red squares > and blue circles: when encountering blue squares, the intended behavior is > undefined. We investigate whether pretrained models are better active > learners, capable of disambiguating between the possible tasks a user may > be trying to specify. Intriguingly, we find that better active learning is > an emergent property of the pretraining process: pretrained models require > up to 5 times fewer labels when using uncertainty-based active learning, > while non-pretrained models see no or even negative benefit. We find these > gains come from an ability to select examples with attributes that > disambiguate the intended behavior, such as rare product categories or > atypical backgrounds. These attributes are far more linearly separable in > pretrained model's representation spaces vs non-pretrained models, > suggesting a possible mechanism for this behavior. > If you're at all interested in building AI systems that can disambiguate concepts or handle uncertainty, please feel free to join. All the best, Quintin Pope Join Zoom Meeting https://oregonstate.zoom.us/j/95843260079?pwd=TzZTN0xPaFZrazRGTElud0J1cnJLU… Password: 961594 Phone Dial-In Information +1 971 247 1195 US (Portland) +1 253 215 8782 US (Tacoma) +1 301 715 8592 US (Washington DC) Meeting ID: 958 4326 0079 All the best, Quintin

1 0

[CoRIS-List] OSU Robotics Seminar 4/22: Kim Ingraham, "Enabling human mobility through personalized robotic control"
by Ross L. Hatton 19 Apr '22

19 Apr '22

Hi All, This week's OSU Robotics seminar will be presented by Kim Ingraham from the University of Washington, on "Enabling human mobility through personalized robotic control". The seminar will be from 10-11am, followed by an 11-11:30 student-only Q&A session. The seminar will be an *in person* event, simulcast over Zoom, and will be immediately followed by a half-hour students-only Q&A session with the speaker. April 22, 10am LINC 268 https://oregonstate.zoom.us/j/93908195212?pwd=cEdoQVdac0JxMUpOSS9xVzZjVG5xU… ----------------- Seminar Title: Enabling human mobility through personalized robotic control Abstract: Assistive robotic technologies—like protheses, exoskeletons, and semi-autonomous powered wheelchairs—have the potential to meaningfully transform human mobility. Yet, despite great strides in the development of wearable assistive robots over the last several decades, people are not widely using these systems in their daily lives. Fundamentally, we do not yet know how to apply robotic assistance to the human body in order to promote meaningful clinical improvements or achieve targeted physiological goals. To this end, my work is focused on the design and experimental evaluation of personalized, adaptive control strategies for assistive robotic devices. In this seminar, I will present work that advances our understanding of how to provide robotic assistance to users outside the laboratory environment. As an example of a personalized control system, I will demonstrate that exoskeleton users can quickly and precisely identify features of robotic assistance that they prefer, and highlight characteristics of user preference that make the design of personalized control systems both compelling and challenging. While many assistive robots support users during walking, I will also share insights into how these methods may be applied to wheeled mobility technologies, specifically for children with disabilities. Together, this work supports my future research goal of designing personalized, adaptive control strategies for wearable assistive robots in order to enable people to meet their goals and achieve full participation in their daily lives. Bio: Kim Ingraham (she/her) is a post-doctoral fellow at the University of Washington in the departments of Rehabilitation Medicine and Mechanical Engineering, and is a member of the multidisciplinary group CREATE (Center for Research and Education on Accessible Technology and Experiences). Kim holds degrees in Mechanical Engineering from the University of Michigan (PhD, 2021) and in Biomedical Engineering from Vanderbilt University (BE, 2012). Prior to beginning graduate school, she worked on an interdisciplinary team of scientists, engineers, and clinicians as a Research Engineer at the Shirley Ryan AbilityLab (2012-2015). Kim is passionate about using assistive wearable robots to advance human mobility in all forms. In her research, she has developed and evaluated physiologically-inspired control systems for a variety of assistive devices, including powered lower-limb prostheses, robotic exoskeletons, and powered wheelchairs for young children with disabilities. Kim is an NSF graduate research fellow, and served as both a graduate student instructor and an engineering teaching consultant at the University of Michigan. =========== Ross L. Hatton Associate Professor, Robotics and Mechanical Engineering Collaborative Robotics and Intelligent Systems Institute Oregon State University coris.oregonstate.edu research.engr.oregonstate.edu/lram/ rosslhatton.com ross.hatton(a)oregonstate.edu

1 0

AI seminar: April 20, 2022
by Mangannavar, Rajesh Devaraddi 19 Apr '22

19 Apr '22

Dear all, Our next AI seminar on *"**Upscaling Nonlinear Reactions in Tissues: Closure via Machine Learning**" *by professor Brian Wood is scheduled to be on April 20th (Tomorrow), 1-2 PM PST. It will be followed by a 30-minute Q&A session by the graduate students. Zoom Link: https://oregonstate.zoom.us/j/93591935144?pwd=YjZaSjBYS0NmNUtjQzBEdzhPeDZ5U… *Upscaling Nonlinear Reactions in Tissues: Closure via Machine Learning* Brian Wood Professor Chemical, Biological, and Environmental Engineering Oregon State University *Abstract:* In this work, we use a combination of formal upscaling and data-driven machine learning for explicitly closing a nonlinear transport and reaction process in a multiscale tissue. The classical effectiveness factor model is used to formulate the macroscale reaction kinetics. We parsimoniously train a multilayer perceptron network using training data generated by direct numerical simulations over thousands of microscale examples. Once trained, the network is applied in an algorithm for numerically solving the upscaled (coarse-grained) differential equation describing mass transport and reaction in two example tissues. The network is described as being explicit in the sense that the network is trained using macroscale concentrations and gradients of concentration as components of the feature space. Network training and solutions to the macroscale transport equations were computed for two different tissues. The two tissue types (brain and liver) exhibit markedly different geometry and spatial scale (cell size and sample size). The upscaled solutions for the average concentration are compared with numerical solutions derived from the microscale concentration fields by a posteriori averaging. There are two outcomes of this work of particular note: 1) we find that the trained network exhibits good generalizability, and it is able to predict the effectiveness factor with high fidelity for realistically-structured tissues despite the significantly different scale and geometry of the two example tissue types; and 2) the approach results in an upscaled PDE with an effectiveness factor that is predicted (implicitly) via the trained neural network. This latter result emphasizes our purposeful connection between conventional averaging methods with the use of machine learning for closure; this contrasts with some machine learning methods for upscaling where the exact form of the macroscale equation remains unknown. *Speaker Bio:* Brian Wood specializes in transport phenomena at multiple scales. His research usually integrates theory, numerical simulation, and experimental data. Systems that include a biological component has been a particular focus, and recent work has involved the multiscale modeling of tissues with closures informed by machine learning. Wood has a significant and evolving interest in parsimoniously using deep neural networks to help “close” nonlinear problems in formal upscaling (e.g., multi-material mechanics, turbulence in fluid mechanics, etc.). The connections between thermodynamics and statistical mechanics is a personal interest; he is actively involved in research to understand how entropy, information, and complexity connect in physical systems. *Please watch this space for future AI Seminars :* * https://eecs.oregonstate.edu/ai-events <https://eecs.oregonstate.edu/ai-events>* Rajesh Mangannavar, Graduate Student Oregon State University ---- AI Seminar Important Reminders: -> For graduate students in the AI program, attendance is strongly encouraged.

1 0

[CoRIS-List] Updated Schedule: Oregon State Robotics Spring 2022 Seminar Series
by Ross L. Hatton 18 Apr '22

18 Apr '22

Hi All, Because a large number of Robotics students will be at the NW Robotics Symposium at the University of Washington on May 13, we are cancelling seminar that day. ICRA practice talks will now be on April 29th; Anna Nickelson has graciously agreed to move her "How to Give a Good Presentation" talk to a new day (to be decided). The full slate of seminars is attached below. Seminars will be from 10-11am, with an 11-11:30 students-only discussion. The seminars will be held in LINC 268 (note that this is different from the room we had in Fall and Winter terms) The Zoom link is https://oregonstate.zoom.us/j/93908195212?pwd=cEdoQVdac0JxMUpOSS9xVzZjVG5xU…. --- 4/1/2022 Cristina Wilson -UPenn GRASP Lab & RAD Lab -Collaborative human-robot teaming -https://www.radlab.us 4/8/2022 RGSA Round Table 4/15/2022 Chris Neider -Starship Technologies -Food delivery robots -https://www.starship.xyz/ 4/22/2022 Kim Ingraham -University of Washington Mechanical Engineering, Rehabilitation Medicine -Biomechanics and wearable robots -https://kim-ingraham.com 4/29/2022 ICRA practice talks 5/6/2022 Shai Revzen -University of Michigan Electrical Engineering -Dynamics, bioinspired, legged locomotion -https://robotics.umich.edu/profile/shai-revzen/ 5/13/2022 No Seminar: NW Robotics Symposium at University of Washington 5/20/2022 Chris Sanchez -OSU Psychology -Human use of computing technology -https://liberalarts.oregonstate.edu/users/christopher-sanchez 5/27/2022 No seminar: ICRA 6/3/2022 Edward Wang -UCSD ECE & Design Lab -Design and health monitoring -https://www.ejaywang.com (Kristen Macuga from OSU Psychology was originally scheduled to speak this term, but has rescheduled for the Fall) =========== Ross L. Hatton Associate Professor, Robotics and Mechanical Engineering Collaborative Robotics and Intelligent Systems Institute Oregon State University coris.oregonstate.edu research.engr.oregonstate.edu/lram/ rosslhatton.com ross.hatton(a)oregonstate.edu

1 0

AI Alignment Reading Group Meeting
by Pope, Quintin 15 Apr '22

15 Apr '22

Hello everyone, I haven't had time to find and analyze a good paper for the reading group. I will, however, still be holding the meeting at 2 PM PST. I've been working on a new perspective on the origin and nature of human values. If you want to read about it, I've made a LessWrong comment <https://www.lesswrong.com/posts/pSdxQfbvC5qBBdehq/?commentId=36PxXrBZvkgWye…> that briefly describes some of the intuition around this new perspective and why I'm excited about it. If that comment intrigued you, I'll be available to discuss more at 2. Please feel free to join! Join Zoom Meeting https://oregonstate.zoom.us/j/95843260079?pwd=TzZTN0xPaFZrazRGTElud0J1cnJLU… Password: 961594 Phone Dial-In Information +1 971 247 1195 US (Portland) +1 253 215 8782 US (Tacoma) +1 301 715 8592 US (Washington DC) Meeting ID: 958 4326 0079 All the best, Quintin

1 0

[CoRIS-List] Reminder: Today, 10AM OSU Robotics Seminar, Starship Robotics
by Ross L. Hatton 15 Apr '22

15 Apr '22

Hi All, This week, the OSU Robotics Seminar will be hosting a talk by Dylan Barrett (replacing Chris Neider) of Starship Robotics (the company behind the food delivery robots on campus). Dylan has launched Starship's delivery services at several large campuses, and manages their operations across a range of campuses with different sizes and needs. He will be presenting on both Starship's operating model and on the technology that makes it possible, and will have a robot with him at the seminar. The seminar will be an *in person* event, simulcast over Zoom, and will be immediately followed by a half-hour students-only Q&A session with the speaker. April 15, 10am LINC 268 https://oregonstate.zoom.us/j/93908195212?pwd=cEdoQVdac0JxMUpOSS9xVzZjVG5xU… =========== Ross L. Hatton Associate Professor, Robotics and Mechanical Engineering Collaborative Robotics and Intelligent Systems Institute Oregon State University coris.oregonstate.edu research.engr.oregonstate.edu/lram/ rosslhatton.com ross.hatton(a)oregonstate.edu

1 0