Jason Morton

Algebraic Statistics June 8-15 2012 at Penn State

Sat, 27 Aug 2011 16:00:00 EDT

We have begun planning for a large Algebraic Statistics conference June 8-15, 2012 at Penn State. The meeting will include tutorials over the weekend followed by a week of invited and contributed talks, and a poster session. Details are on the conference website.

Mathematical foundations of deep learning workshop June 21-22

Mon, 27 Jun 2011 16:00:00 EDT

As part of an effort to bring mathematicians and computer scientists working in the area together to explore and strengthen the mathematical foundations of deep learning, I organized organizing a two-day workshop on the topic. The purpose of the workshop was to discuss strategies for improving these foundations, including but not limited to

formal statements and proofs of the "deep architectures are better" hypothesis,
mathematical strategies for a priori comparison of architectures for learning tasks,
questions about deep belief networks that can be translated to problems in algebraic and tropical geometry, such as identifiability and singular learning theory,
inference functions, encodings, and circuit complexity, and
how all of the above relates to learning algorithms.

Thanks to the wonderful participants, the workshop led to a lot of interesting new ideas and research directions.

DARPA-NCI Cancer Prediction, Prognosis, and Prevention (CAP3) Workshop 4/25-26/11

Wed, 27 Apr 2011 16:00:00 EDT

I just attended the DARPA-NCI Cancer Prediction, Prognosis, and Prevention (CAP3) Workshop in Bethesda. The workshop brought together some outstanding cancer researchers and representatives of NCI and NIH with some mathematics and CS folks from the DARPA side to look for ways to collaborate on cancer diagnosis and treatment. An example is trying to predict patient response to a drug based on a heterogeneous dataset of biomarkers, sequence data, etc. Apparently, cancer has been shockingly resistant to translating biomarker studies that work as correlates or predictors in a research setting to something that works clinically. The analytics camp has been trying to come up with general-purpose tools, especially for predicting failure modes in complex systems (example), and the hope is that these might be brought to bear on such prediction tasks.

ICERM Topical Workshop on Mathematical Aspects of P versus NP and its Variants, August 1-5, 2011

Sun, 24 Apr 2011 08:00:00 EDT

I'll be speaking at the inaugural ICERM topical workshop, on Mathematical Aspects of P versus NP and its Variants, which meets August 1-5, 2011. It is organized by Saugata Basu, JM Landsberg, and J Maurice Rojas. From the description:

This workshop will bring together computer scientists and mathematicians to examine the P v. NP problem and its variants from the perspectives of algebra, geometry, and number theory, and to introduce the mathematical aspects of these questions to a larger audience. Diverse researchers working on different aspects of these problems will clarify connections between different approaches.

There will be two main topics: Analogues of P v. NP (e.g., Valiant's conjectures, the Mulmuley-Sohoni Conjecture, the BSS model, and other computational models); and Algebraic, Number Theoretic, and Geometric Aspects of P v. NP (e.g., Holographic algorithms, characterizations of NP in terms of sheaf cohomology, sparse polynomials, and other arithmetic approaches).