BAIR/CPAR/BDD Internal Weekly Seminar

The Berkeley Artificial Intelligence Research Lab co-hosts a weekly internal seminar series with the CITRIS People and Robots Initiative and the Berkeley DeepDrive. The seminars are every Tuesday morning, from 11:10A-12P, and are open to BAIR faculty, students, and sponsors. Talks will take place via Zoom. For any questions, please email bair-admin@berkeley.edu.

Fall 2020 Schedule

Date	Speaker 1: 11:10 AM -12:00 PM or 11:10-11:30 AM	Speaker 2: N/A or  11:30-11:50 AM
Sep 1	Alvin Wan: What Explainable AI Fails to Explain (and how we fix that)	Amir Gholami, Zhewei Yao: ADAHESSIAN: An Adaptive Second Order Optimizer for Machine Learning
Sep 8	Nikita Kitaev: Is Unstructured Computation All You Need for Natural Language Processing?	Lucy Li: Content Analysis of Textbooks via Natural Language Processing: Findings on Gender, Race, and Ethnicity in Texas U.S. History Textbooks
Sep 15	Georgios Pavlakos: Learning to Reconstruct 3D Humans	N/A
Sep 22	Lerrel Pinto: Rich Robotic Supervision through Cheap Human Demonstrations	N/A
Sep 29	Philippe Laban: Text Summarization Without The Summaries	Katie Stasaski: More Diverse Dialogue Datasets via Diversity-Informed Data Collection
Oct 6	Aldo Pacchiano:  Learning to Score Behaviors for Guided Policy Optimization	N/A
Oct 13	Paria Rashidinejad: Learning to Predict in Unknown Dynamical Systems with Long-Term Memory	Zhen Dong: Efficient Neural Networks Through Systematic Quantization
Oct 20	Eugene Vinitsky: Optimizing Energy Efficiency of Traffic at Scale via Multi-agent Deep RL	Anastasios Angelopoulos: Uncertainty for Black Box Models
Oct 27	Misha Laskin: TBD	N/A
Nov 3	Wei Zhan: TBD	Liting Sun: TBD
Nov 10	Daniel Brown: Safe and Efficient Imitation Learning	N/A
Nov 17	Edward Kim: Scenic: Dynamic Scenario Description Language for Autonomous Systems	Kimin Lee: Ensemble methods in reinforcement learning
Nov 24	Xin Wang: TBD	N/A
Dec 1	Huijuan Xu: TBD	N/A

---

Spring 2020 Schedule

Date	Speaker 1: 3:10-4:00 or 3:10-3:30	Speaker 2: N/A or 3:30-3:50
Jan. 24	Dylan Hadfield-Menell: The Principal-Agent Value Alignment Problem	N/A
Jan. 31	Somil Bansal: Safe and Data-efficient Learning for Physical Systems	N/A
Feb. 7	Nick Antipa: Lensless Computational Imaging: Seeing More with Less	N/A
Feb. 14	Ruoxi Jia: Towards a Responsible Data Economy: Fairness, Privacy, and Security	N/A
Feb. 21	Chandan Singh: Interpreting and Improving Neural Networks via Disentangled Attributions	N/A
Feb. 28	Alexei (Alyosha) Efros: Image Manipulation… And Ways to Detect It	N/A
Mar. 6	Brijen Thananjeyan and Ashwin Balakrishna:  Safety Augmented Value Estimation from Demonstrations (SAVED): Safe Deep Model-Based RL for Sparse Cost Robotic Tasks	Jeff Ichnowski: Fog-Robotics Serverless and Deep Grasp-Optimized Robot Motion Planning
Mar. 13	Aravind Srinivas: Self-Supervised Visual Representation Learning	N/A
Mar. 20	Ronghang Hu: Structured Models for Vision-and-Language Reasoning	N/A
Mar. 27	No Seminar	N/A
Apr. 3	Misha Laskin: Improving Reinforcement Learning with Unsupervised and Self-Supervised Learning	N/A
Apr. 10	Dequan Wang: Object-Centric Representation for Perception, Prediction, and Planning	N/A
Apr. 17	Coline Devin:	Learning with Modularity and Compositionality for Robotics
Apr. 24	Cecilia Zhang: Bringing Cinema Quality Rendering into Casual Photos and Videos.	N/A
May 1	Sylvia Herbert: Safe Real-World Autonomy in Uncertain and Unstructured Environments	N/A
May 8	Sasha Sax: Robust Learning Through Cross-Task Consistency	N/A
May 15	Carlos Florensa: What Supervision Scales? Practical Learning Through Interaction	N/A

---

Spring 2019 Schedule

Date	Speaker 1: 3:10-4:00 or 3:10-3:30	Speaker 2: N/A or 3:30-3:50
Jan. 18	Eric Jonas: Structured Prediction via Machine Learning for Inverse Problems	N/A
Jan. 25	Dinesh Jayaraman: Towards Embodied Visual Intelligence	N/A
Feb. 1	Ke Li: Advances in Machine Learning: Learning to Optimize, Generative Modelling and Nearest Neighbour Search	N/A
Feb. 8	Deepak Pathak: Self-Directed Learning	N/A
Feb. 15	Samaneh Azadi: Rectify the GAN Generator Distribution by Rejecting Bad Samples	Sasha Sax: On Perception for Robotics: Mid-level Visual Representations Improve Generalization and Sample Complexity for Learning Active Tasks
Feb. 22	Jaime Fisac: Resilient Safety Assurance for Robotic Systems: Staying Safe Even When Models Are Wrong	N/A
Mar. 1	Fisher Yu: Towards Human-Level Recognition via Contextual, Dynamic, and Predictive Representations	N/A
Mar. 8	Andrew Owens: Sight and Sound	N/A
Mar. 15	Daniel Fried: Pragmatic Models for Generating and Following Grounded Instructions	Nikita Kitaev: Syntactic Parsing with Self-Attention
Mar. 22	**Oriol Vinyals: AlphaStar: Mastering the Real-Time Strategy Game StarCraft II (**This will be held 11am-12pm in 250 SDH**)	N/A
Mar. 22	Jacob Steinhardt: What Makes Neural Networks (Non-)Robust?	N/A
Mar. 29	Holiday	Holiday
Apr. 5	Alex Lee: Visual Dynamics Models for Robotic Planning and Control	N/A
Apr. 12	Lisa Anne Hendricks: Visual Understanding through Natural Language	N/A
Apr. 19	Evan Shelhamer: Blurring the Line between Structure and Learning for Adaptive Local Recognition	N/A
Apr. 26	Yian Ma: Bridging MCMC and Optimization	N/A
May 3	Sandy Huang: Optimizing for Robot Transparency	N/A
May 10	Jeffrey Regier: Statistical Inference for Cataloging the Visible Universe	N/A

---

Fall 2018 Schedule

Date	Speaker 1: 3:10-4:00 or 3:10-3:30	Speaker 2: N/A or 3:30-3:50
Aug. 24	Jaime Fisac, Andrea Bajcsy, Sylvia Herbert: Probabilistically Safe Robot Planning with Confidence-Based Human Predictions	N/A
Aug. 31	Aviv Tamar: Learning Representations for Planning	N/A
Sept. 7	Chi Jin: Is Q-learning Provably Efficient?	N/A
Sept. 14	Wojciech Zaremba: Learning dexterity	N/A
Sept. 21	Tuomas Haarnoja: Acquiring Diverse Robot Skills via Maximum Entropy Reinforcement Learning	N/A
Sept. 28	Professor Ruzena Bajcsy: Data Driven vs. Model Driven Analysis of Human Ability for Human Robot Interaction	N/A
Oct. 5	Ke Li: Implicit Maximum Likelihood Estimation	N/A
Oct. 12	Roy Fox: Multi-Task Hierarchical Imitation Learning of Robot Skills	N/A
Oct. 19	Ankush Desai: DRONA: A Framework for Programming Safe Robotics Systems	N/A
Oct. 26	Ajay Tanwani: A Fog Robotics Approach to Large Scale Robot Learning	Laura Hallock: Human Muscle Force Modeling for Enhanced Assistive Device Control
Nov. 2	Abhishek Gupta: Unsupervised (Meta) RL	Anja Rohrbach: Diagnosing and correcting bias in captioning models
Nov. 9	Holiday	Holiday
Nov. 16	Daniel Fried: Pragmatic Models for Generating and Following Grounded Instructions	Nikita Kitaev: Syntactic Parsing with Self-Attention
Nov. 23	Holiday	Holiday
Dec. 7	Jiantao Jiao: Deconstructing Generative Adversarial Networks	N/A

---

Spring 2018 Schedule

Date	Speaker 1: 3:10-4:00 or 3:10-3:30	Speaker 2: N/A or 3:30-3:50
Jan. 19	Pulkit Agrawal: Continually Evolving Machines: Learning by Experimenting	N/A
Jan. 26	Sanjay Krishnan: Dirty Data, Robotics, and Artificial Intelligence	N/A
Feb. 2	David Fouhey: Towards a 3D World of Interaction	N/A
Feb. 9	Saurabh Gupta: Visual Perception and Navigation in 3D Scenes	N/A
Feb. 16	Jacob Andreas: Learning from Language	N/A
Feb. 23	Chelsea Finn: Generalization and Self-Supervision in Deep Robotic Learning	N/A
Mar. 2	Jennifer Listgarten: Where genetics and biology meet machine learning	N/A
Mar. 9	Yi Ma: Low-dimensional Structures and Deep Models for High-dimensional Data	TBD
Mar. 16	Cathy Wu: Mixed-autonomy mobility: scalable learning and optimization	N/A
Mar. 23	Xiang Cheng: Langevin MCMC as gradient flow over the probability space	Eric Tzeng and Andreea Bobu: Domain Adaptation for Fixed and Continuously Varying Domains
Apr. 6	Shubham Tulsiani: Learning Single-view 3D Reconstruction of Objects and Scenes	TBD
Apr. 13	Tinghui Zhou: Beyond Direct Supervision: Visual Learning via Data-driven Consistency Consistency (Location: SWARM Lab 490 Cory)	N/A
Apr. 20	Richard Zhang: Image Synthesis for Self-Supervised Representation Learning	N/A
Apr. 27	Reza Abbasi-Asl: Structural compression of Convolutional Neural Networks	Fereshteh Sadeghi: Sim2Real Viewpoint Invariant Visual Servoing by Recurrent Control

---

Fall 2017 Schedule

Date	Speaker 1: 3:10-3:40	Speaker 2: 3:40-4:10
Aug. 25	Jitendra Malik: What have we learned from datasets in computer vision?	Angjoo Kanazawa: Single-View 3D Reconstruction of Deformable Objects like Animals and People
Sept. 1	Chang Liu: Neural Network-based Graph Embedding for Cross-Platform Binary Code Similarity Detection	Claire Tomlin: Safe Learning
Sept. 8	Andrew Critch: Open-source game theory is weird	Anna Rohrbach: Generation and grounding of natural language descriptions for visual data
Sept. 15	Bo Li: Secure learning in adversarial environments	Ching-Yao Chan: Safety of Automated Driving Systems (ADS) and AI/ML – A Dialogue
Sept. 22	Alex Anderson: The High-Dimensional Geometry of Binary Neural Networks	Jeff Mahler: Learning Deep Policies for Robot Bin Picking using Discrete-Event Simulation of Robust Grasping Sequences
Sept. 29	Fisher Yu: Towards Universal Representation for Image Recognition	Max Rabinovich: Abstract Syntax Networks for Code Generation and Semantic Parsing
Oct. 6	Michael Laskey: Learning Home Robotics Manipulation Tasks from Remote Supervision	Pulkit Agrawal: Continually Evolving Agents: Curiosity & Experimentation
Oct. 13	Ron Fearing: Dextrous Locomotion	Sylvia Herbert and David Fridovich: Planning, Fast and Slow with FaSTrack: A Framework for Adaptive Real-Time Safe Trajectory Planning
Oct. 20	Emrah Bostan: Learning Convex Regularizers for Optimal Bayesian Denoising	Somil Bansal: Overcoming Model Bias in Model-based Learning
Oct. 27	Roberto Calandra	Aaditya Ramdas: Sequential testing and online false discovery rate control
Nov. 3	Roy Fox: Discovery of Hierarchical Structures for Robot Learning and Neural Programming	Leila Wehbe: Modeling brain responses to natural language stimuli
Nov. 10	Holiday	Holiday
Nov. 17	Jacob Andreas: Learning with Latent Language	Niladri Chatterji: Alternating minimization for dictionary learning with random initialization
Nov. 24	Holiday	Holiday
Dec. 1	Eric Jonas: Could a neuroscientist understand a microprocessor	Tuomas Haarnoja: Soft Q-Learning

---

Spring 2017 Schedule

Date	Speaker 1: 3:10-3:40	Speaker 2: 3:40-4:10
Jan. 20	Jack Gallant: Melding neuroscience and computer science	Javad Lavaei: On optimization theory, numerical algorithms and machine learning for nationwide energy systems
Jan. 27	Trevor Darrell: Adaptive Learning of Driving Models from Large-scale Video Datasets	Coline Devin: Transfer learning for robotics
Feb. 3	Phillip Isola: Image-to-Image Translation with Conditional Adversarial Networks	Anil Aswani: Human Modeling Using Inverse Optimization
Feb. 10	Zeynep Akata: Generating Fine-Grained Visual Explanations and Realistic Images	Deirdre Mulligan: Hand-offs in AI: Functional Fidelity, Values, and Governance
Feb. 17	(3:10- 3:25) Jaime Fernandez Fisac: Generating plans that predict themselves/ (3:25-3:40) Aaron Bestick: Implicitly Assisting Humans to Choose Good Grasps in Robot to Human Handovers	William Guss: WaveLayers: Neural Topology Inspired by Topology
Feb. 24	Laurent El Ghaoui: Safe feature elimination in sparse learning	Laura Waller: Computational Microscopy with sparsity
Mar. 3	Chi Jin: How to Escape Saddle Points Efficiently	Horia Mania: Universality of Mallows’ and degeneracy of Kendall’s kernels for rankings
Mar. 10	Matthew Matl: Automated Grasp Transfer with Mesh Segmentation and Point Cloud Registration	Paul Grigas: An Extended Frank-Wolfe Method with “In-Face” Directions, and its Application to Low-Rank Matrix Completion
Mar. 17	Roy Fox: Multi-Level Discovery of Deep Options	Parvez Ahammad: Beyond navigation metrics: Teaching computers how users perceive web application performance
Mar. 24	Ke Li: Learning to Optimize and Fast k-Nearest Neighbour Search	Jessica Hamrick: Metacontrol for Imagination-Based Optimization
Apr. 7	Deepak Pathak: Exploring Four Axes of Self-Supervision	Saurabh Gupta: Cognitive Mapping and Planning for Visual Navigation
Apr. 14	Jeff Mahler: Dex-Net: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics	Judy Hoffman: A General Framework for Domain Adversarial Learning
Apr. 21	Fereshteh Sadeghi: Sim2Real Collision Avoidance for Indoor Navigation of Mobile Robots via Deep Reinforcement Learning	Evan Shelhamer: Loss is its own Reward: Self-Supervision for Reinforcement Learning
Apr. 28	Roberto Calandra: Robustness in Multi-objective Bayesian optimization	Richard Zhang: Cross-Channel Visual Prediction
May 5	Abhishek Gupta: Imitation Learning for Dexterous Manipulation and Transfer in Reinforcement Learning	Tinghui Zhou: 3D Visual Synthesis and Understanding from 2D Views
May 12	Karl Zipser: New Trajectories in the Autonomous Model Car Project	Daniel Drew: Silent Swarms: Flying Microrobots Using Atmospheric Ion Thrusters

---

Fall 2016 Schedule

Date	Speaker 1: 3:10-3:40	Speaker 2: 3:40-4:10
Aug. 26	Ken Goldberg: Deep Dexterity	Sanjay Krishnan: Inverse Reinforcement Learning For Sequential Robotic Tasks
Sep 2	Anca Dragan: How Robots Influence Our Actions	Jeff Donahue: Adversarial Feature Learning
Sep 9	Robert Nishihara & Philipp Moritz: Distributed Machine Learning with Ray	John DeNero: Interactive machine translation
Sep 16	Joseph Gonzalez: Prediction Serving and RISE	Kevin Jamieson: Bayesian Optimization and other bad ideas for hyperparameter tuning
Sep 23	Stuart Russell: Human-Compatible AI	Marcus Rohrbach: Explain and Answer: Intelligent systems which can communicate about what they see.
Sep 30	Richard Zhang: Colorful Image Colorization	Virginia Smith: A General Framework for Communication-Efficient Distributed Optimization
Oct 7	Gregory Kahn: Learning Control Policies for Partially Observable Safety-Critical Systems	John Canny: Accountable Deep Networks
Oct 14	Ruzena Bajcsy: Individualized Human Models for Cyberphysical Interaction	Jacob Andreas: Neural module networks
Oct 21	Dave Moore: Bayesian seismic monitoring from raw waveforms	Greg Durrett: Data-Driven Text Analysis with Joint Models
Oct 28	Jun-Yan Zhu: Visual Manipulation and Synthesis on the Natural Image Manifold	Dylan Hadfield-Menell: The Off-Switch
Nov 4	Bin Yu: Artificial neurons meet real neurons: building stable interpretations of V4 neurons from CNN+regression models	Pulkit Agrawal: Forecasting from Pixels: Intuitive Physics and Intuitive Behavior
Dec 02	Chelsea Finn: Adversarial Inverse Reinforcement Learning	Aviv Tamar: Deep Policy Representations based on a Planning Computation
Dec 09	Michael Laskey: A Human-Centric Approach to Deep Robotic Learning from Demonstrations.	Michael Oliver: Using artificial neural networks to model visual neurons
Dec 16	Lisa Anne Hendricks: Localizing Moments in Video with Natural Language	Andrew Owens: Learning visual models from paired audio-visual examples