Venue
The meeting will be held in the Mong Auditorium located in the Engineering VI
building on the UCLA campus.
Travel Information
We recommend arranging air travel through LAX. For directions to UCLA from LAX,
check this link and follow directions from the 405 South freeway.
There are direct flights from
Baltimore (Alaska Airlines, American Airlines, United Airlines),
Washington (Alaska Airlines, American Airlines, United Airlines),
San Francisco (American Airlines, Delta Airlines, United Airlines) or
London (United Airlines, American Airlines, British Airways).
Accommodation
Lodging is available at the Luskin Conference Center, directly across the street from the meeting venue. The School of Engineering discounted rate is $245/night.
Government employees can request government rate by calling the reservation desk (1-855-522-8252). Other recommended lodging options are the Kimpton Palomar Hotel, W Los Angeles, and the Hilgard House. We would recommend avoiding the Hotel Angeleno and the Luxe Sunset.
Schedule
The meeting is tentatively scheduled to go from 8:00 AM to 6:30 PM on the 26th, and from 8:00 AM to 2:00 PM on the 27th. A complete (tentative) schedule is given below.
A tentative list of scheduled posters for the Thursday 5:25 PM session is as follows
| September 26 | Part I: Review and Progress Report (closed session) | |
|---|---|---|
| 08:00 -- 08:30 | Registration, light breakfast (coffee, bagels) | |
| 08:30 -- 08:50 | Overview of the program | Hamid Krim (ARO) and Paul Thomas (DSTL) |
| 08:50 -- 09:20 | Overview of the project | Rene Vidal (JHU) |
| 09:20 -- 09:35 | Aims 1 & 5: Characterization of Semantic Information Content; Semantic Information Fusion | Rene Vidal (JHU) |
| 09:40 -- 10:05 | Aim 2: Impact of Data Transformations and Processing on Information Content | Stefano Soatto (UCLA) |
| 10:10 -- 10:40 | Coffee Break | |
| 10:40 -- 11:15 | Aim 3: Optimization for Learning Most Informative Data Representations | Jason Lee (Princeton) |
| 11:20 -- 11:45 | Aim 4: Characterizing Uncertainty in Multimodal Information Representations | Arnaud Doucet (Oxford) |
| 11:50 -- 12:05 | Aim 6: Validation and Integration | Miroslaw Bober (Surrey) |
| 12:10 -- 01:10 | Lunch (free-form: many options at Ackermann Union) | |
| Part II: Unforeseen Breakthroughs and Wild Ideas (open session) | ||
| 01:10 -- 01:30 | Making the black box effective | Emmanuel Candes (Stanford) |
| 01:35 -- 01:55 | Information and generalization | John Shawe-Taylor (UCL) |
| 02:00 -- 02:20 | The dynamics of differential learning | Stefano Soatto (UCLA) |
| 02:25 -- 02:45 | Exact simulation for uncertainty quantification | Arnaud Doucet (Oxford) |
| 02:50 -- 03:20 | Coffee Break | |
| 03:20 -- 03:40 | On Gradient-Based Methods for Finding Game-Theoretic Equilibria | Michael Jordan (Berkeley) |
| 03:45 -- 04:05 | Generative and discriminative domain adaptation and generalization | Rama Chellappa (UMD) |
| 04:10 -- 04:30 | The changing face of the curse of dimensionality | Josef Kittler (Surrey) |
| 04:35 -- 04:55 | ACTNET: Learning aggregations for object recognition | Miroslaw Bober (Surrey) |
| 05:00 -- 05:20 | Linearly Convergent Non-Smooth Non-Convex Optimization for Robust Subspace Learning | Rene Vidal (JHU) |
| 05:25 -- 06:25 | Poster session | |
| 06:30 | Dinner for PIs and Program Officers | Plateia Restaurant (across the street) |
| September 27 | Part III: Interactive and Next Generation (open session) | |
| 08:00 -- 08:30 | Registration, light breakfast (coffee, bagels) | |
| 08:30 -- 08:40 | Contextual Information Separation for Moving Object Segmentation | Yanchao Yang (UCLA) |
| 08:40 -- 08:50 | PAC-Bayes generalisation bounds for binary-activated neural networks | Benjamin Guedj (UCL) |
| 08:55 -- 09:05 | Geometric Integrators and Splitting methods in Optimization | Guilherme Franca (JHU) |
| 09:05 -- 09:15 | Langevin Monte Carlo Without Smoothness | Jelena Diakonikolas (Berkeley) |
| 09:20 -- 09:30 | Dual space preconditioning for gradient descent | Daniel Paulin (Oxford) |
| 09:30 -- 09:40 | Alessandro Barp (Cambridge) | |
| 09:40 -- 09:50 | EntropicGANs meet VAEs: A statistical approach to compute sample likelihoods in GANs | Yogesh Balaji (UMD) |
| 10:00 -- 10:30 | Coffee Break, discussion | |
| 10:30 -- 11:30 | Workshop Session: Unifying Use Case | Paul Thomas (DSTL) |
| 11:30 -- 12:15 | Government Closed Session | |
| MURI PIs & Student Planning Session | ||
| 12:15 -- 01:15 | Lunch (free-form) | |
| 01:15 -- 01:45 | Discussion, Program Feedback | ARO, DLST, PIs |
| 01:45 -- 02:00 | Adjourn |
| 1 | Conditional Prior Network | Yanchao Yang (UCLA) |
|---|---|---|
| 2 | Unsupervised Domain Adaptation by Regularized Conditional Alignment | Safa Cicek (UCLA) |
| 3 | Zero-shot Learning with the Isoperimetric Loss | Shay Deutsch (UCLA) |
| 4 | Critical Regularization Periods in Deep Networks | Aditya Golaktar (UCLA) |
| 5 | GeoNets | Tong He (UCLA) |
| 6 | Geo-supervised Scene Modeling | Alex Wong (UCLA) |
| 7 | Exploring Dataset Fusion. A comparison of different strategies: From graphical models to multi-task learning | Maria Perez (UCL) |
| 8 | Dual space optimization methods | Daniel Paulin (Oxford) |
| 9 | Fitting Diversified Data Distributions with Supervised GANs | Ilya Kavalerov (UMD) |
| 10 | Conformal Symplectic and Relativistic Optimization | Guilherme Franca (JHU) |
| 11 | Gradient Flow and Accelerated Proximal Splitting Methods | Guilherme Franca (JHU) |
| 12 | A Hamiltonian Perspective on Generalized Momentum-Based Methods in Optimization | Jelena Diakonikolas (Berkeley) |
| 13 | HMC by Symplectic Reduction | Alessandro Barp (Cambridge) |