July 2019 - Candle-aesp - mailman3.cels.anl.gov

Re: [Candle-aesp] AI assesses long-term mortality with single x-ray
by Brettin, Thomas S. 23 Jul '19

23 Jul '19

Would be cool if the model were made public. I’d ask my doctor for a copy of my most recent chest xray. On 7/22/19, 10:05 PM, "Candle-aesp on behalf of Stevens, Rick L. via Candle-aesp" <candle-aesp-bounces(a)lists.cels.anl.gov on behalf of candle-aesp(a)lists.cels.anl.gov> wrote: https://www.radiologybusiness.com/topics/artificial-intelligence/ai-deep-le… Sent from my iPhone -- Candle-aesp mailing list Candle-aesp(a)lists.cels.anl.gov https://lists.cels.anl.gov/mailman/listinfo/candle-aesp

1 0

AI assesses long-term mortality with single x-ray
by Stevens, Rick L. 22 Jul '19

22 Jul '19

https://www.radiologybusiness.com/topics/artificial-intelligence/ai-deep-le… Sent from my iPhone

1 0

Fwd: [Aurora-esp-team-members] Aurora Programming Workshop 17-19 September 2019
by Stevens, Rick L. 17 Jul '19

17 Jul '19

We should probably send a few. Sent from my iPhone Begin forwarded message: From: ALCF Early Science Program via Aurora-esp-team-members <aurora-esp-team-members(a)lists.alcf.anl.gov<mailto:[email protected]>> Date: July 17, 2019 at 19:45:46 CDT To: "aurora-esp-team-members(a)alcf.anl.gov<mailto:[email protected]>" <aurora-esp-team-members(a)alcf.anl.gov<mailto:[email protected]>> Cc: "Knight, Christopher J." <knightc(a)anl.gov<mailto:[email protected]>>, "Parker, Scott J." <sparker(a)alcf.anl.gov<mailto:[email protected]>>, "Kumaran, Kalyan" <kumaran(a)anl.gov<mailto:[email protected]>> Subject: [Aurora-esp-team-members] Aurora Programming Workshop 17-19 September 2019 Reply-To: ALCF Early Science Program <earlyscience(a)alcf.anl.gov<mailto:[email protected]>> Dear ESP PIs and teams, From Sept. 17-19, 2019, the ALCF will be holding a workshop with a focus on applications and software development for the Aurora system. The workshop is by INVITATION ONLY and will take place at Argonne National Laboratory. While the workshop is being organized for Exascale Computing Project (ECP) developers, the ALCF would like to invite Aurora ESP developers as space is available. Please consider who on your team you might send. Ideal attendees are key developers who are working on moving applications to Aurora’s architecture, and one or two people for a given ESP project would typically be the limit. The workshop will focus specifically on programming the Gen9 systems using our early software development kit (SDK) from Intel. There will be some focus on libraries such as MKL, but not much on ML/DL frameworks or complex workflows; emphasis will be on compiled code. The draft agenda is on the workshop website<https://www.alcf.anl.gov/workshops/aurora-programming-workshop>, where you’ll also find a link for registering; here is the current state of the draft agenda: AGENDA Meeting Location: Argonne National Laboratory, TCS Conference Center Room 1416 Registration Deadline: August 2, 2019 Your attendance is not approved until confirmed by email from ALCF staff. It is not advised to make any travel arrangements until you receive a confirmation. Dates: September 17-19, 2019 · High-level overview of the Aurora system · Presentations on Intel’s Aurora programming models, including OpenMP, SYCL/DPC++, OpenCL · Presentations on open programming models, including Kokkos and Raja · Presentations on the development hardware · Presentations on the Aurora software development kit (supported programming models, compilers, libraries, tools, etc) · Lightning talks on programming progress, best practices, and lessons learned from attendees · Hands-on sessions working with the latest Aurora software development kit This is a good opportunity to make progress on porting to Intel GPUs. Note: the registration page mentions a requirement that an attendee’s institution have a CNDA signed with Intel. The institutional RSNDAs we have signed for the ESP, and being listed the associated MPA, are more than sufficient to meet this requirement. Best regards, Tim Williams, Venkat Vishwanath, Ramesh Balakrishnan [cid:[email protected]] -- Aurora-esp-team-members mailing list Aurora-esp-team-members(a)lists.alcf.anl.gov<mailto:[email protected]> https://lists.alcf.anl.gov/mailman/listinfo/aurora-esp-team-members

1 0

definitely take a look at this one.
by Stevens, Rick L. 11 Jul '19

11 Jul '19

https://arxiv.org/pdf/1906.05221.pdf

1 0

FW: Call for Paper -- The 3rd Deep Learning on Supercomputers Workshop co-located with SC'19 in Denver, CO
by Brettin, Thomas S. 08 Jul '19

08 Jul '19

FYI. From: Zhao Zhang <zzhang(a)tacc.utexas.edu> Date: Monday, July 8, 2019 at 11:59 AM To: Ian Foster <foster(a)uchicago.edu>, Vali Codreanu <valeriu.codreanu(a)surfsara.nl>, Weijia Xu <xwj(a)tacc.utexas.edu> Subject: Call for Paper -- The 3rd Deep Learning on Supercomputers Workshop co-located with SC'19 in Denver, CO Dear deep learning practitioners, We sincerely invite you and your colleagues to submit to the 3rd Deep Learning on Supercomputers Workshop (https://dlonsc19.github.io/) Please circulate this call to your colleagues who work on the topic. The Deep Learning (DL) on Supercomputers workshop will be held with the SC’19 conference in Denver, CO, on Nov 17th, 2019. This third workshop in the Deep Learning on Supercomputers series provides a forum for practitioners working on any and all aspects of DL for scientific research in the High Performance Computing (HPC) context to present their latest research results. The general theme of this workshop series is the intersection of DL and HPC. Its scope encompasses application development in scientific scenarios using HPC platforms; DL methods applied to numerical simulation; fundamental algorithms, enhanced procedures, and software development methods to enable scalable training and inference; hardware changes with impact on future supercomputer design; and machine deployment, performance evaluation, and reproducibility practices for DL applications, with an emphasis on scientific usage. Topics include but are not limited to: · Emerging scientific applications driven by DL methods · Novel interactions between DL and traditional numerical simulation · Effectiveness and limitations of DL methods in scientific research · Algorithms and procedures to enhance reproducibility of scientific DL applications · Data management through the life cycle of scientific DL applications · General algorithms and procedures for efficient and scalable DL training · General algorithms and systems for large scale model serving for scientific use cases · New software, and enhancements to existing software, for scalable DL · DL communication optimization at scale · I/O optimization for DL at scale · Hardware (processors, accelerators, memory hierarchy, interconnect) changes with impact on deep learning in the HPC context · DL performance evaluation and analysis on deployed systems · DL performance modeling and tuning of DL on supercomputers · DL benchmarks on supercomputers · As part of the reproducibility initiative, the workshop requires authors to provide information such as the algorithms, software releases, datasets, and hardware configurations used. For performance evaluation studies, we will encourage authors to use well-known benchmarks or applications with open accessible datasets: for example, MLPerf and ResNet-50 with the ImageNet-1K dataset. For questions, please contact (DLSuperWorkshop19(a)gmail.com) Import Dates (Tentative) · Papers due: Sep 1st, 2019 · Acceptance notification: Sep 16th, 2019 · Camera ready: Oct 1st, 2019 · Workshop date: Nov 17th, 2019 Paper Submission Authors are invited to submit unpublished, original work with a minimum of 6 pages in IEEE conference format (https://www.ieee.org/conferences/publishing/templates.html) and submitted using Linklings (https://submissions.supercomputing.org/?page=Submit&id=SC19WorkshopDeepLear…) (login required). IEEE TCHPC conditionally agrees to publish accepted papers. Organizing Committee · Valeriu Codreanu (co-chair), SURFsara, Netherlands · Ian Foster (co-chair), UChicago & ANL, USA · Zhao Zhang (co-chair), TACC, USA · Weijia Xu (proceeding chair), TACC, USA · Ahmed Al-Jarro, Fujitsu Laboratories of Europe, UK · Takuya Akiba, Preferred Networks, Japan · Thomas S. Brettin, ANL, USA · Erich Elsen, DeepMind, USA · Steve Farrell, LBNL, USA · Song Feng, IBM Research, USA · Boris Ginsburg, Nvidia, USA · Torsten Hoefler, ETH, Switzerland · Jessy Li, UT Austin, USA · Zhengchun Liu, ANL, USA · Peter Messmer, Nvidia, USA · Damian Podareanu, SURFsara, Netherlands · Simon Portegies Zwart, Leiden Observatory, Netherlands · Qifan Pu, Google, USA · Judy Qiu, Indiana University, USA · Arvind Ramanathan, ANL, USA · Vikram Saletore, Intel, USA · Mikhail E. Smorkalov, Intel, Russia · Rob Schreiber, Cerebras, USA · Dan Stanzione, TACC, USA · Rick Stevens, UChicago & ANL, USA · Wei Tan, Citadel, USA · Jordi Torres, Barcelona Supercomputing Center, Spain · Daniela Ushizima, LBNL, USA · Sofia Vallecorsa , CERN, Switzerland · David Walling, TACC, USA · Markus Weimer, Microsoft, USA · Kathy Yelick, UC Berkeley & LBNL, USA For previous workshops, please check 1st Deep Learning on Supercomputers Workshop in SC'18, https://www.tacc.utexas.edu/workshop/2018/deep-learning 2nd Deep Learning for Science Workshop in ISC'19, https://dlonsc.github.io/

1 0

NCIP Hub - Group: Fifth Computational Approaches for Cancer Workshop
by Stevens, Rick L. 01 Jul '19

01 Jul '19

We definitely should plan on submitting a number of abstracts to this workshop. Its focused on Drug Discover and AI in Cancer. So our AESP project is perfect. as are a few related topics .. Abstracts are due in Sept. https://nciphub.org/groups/cafcw19 Fifth Computational Approaches for Cancer Workshop (CAFCW-2019) Held in conjunction with the SC19: The International Conference on High Performance Computing, Networking, Storage and Analysis Sunday November 17, 2019, Denver, Colorado, USA 8:30AM – 5:00 PM CALL FOR ABSTRACTS <https://nciphub.org/groups/cafcw19/call_for_abstracts> New computational opportunities and challenges have emerged within the cancer research and clinical application areas as the size, number, variety and complexity of cancer datasets have grown in recent years. Simultaneously, advances in computational capabilities such as deep learning present growing opportunities to transform cancer data into cancer knowledge. Such opportunities to impact cancer computationally are underscored in the 2016 US Twenty-first Century Cures Act and international efforts such as the ITCC-P4 focusing on development treatments for pediatric cancer. The workshop focuses on bringing together interested individuals ranging from clinicians, mathematicians, data scientists, computational scientists, engineers, developers, leaders and others with an interest in advancing the use of computation at all levels to better understand, diagnose, treat and prevent cancer. As an interdisciplinary workshop, the cross-disciplinary sharing of insight and challenges fosters collaborations and future innovations to accelerate the progress in computationally and data driven cancer research and clinical applications. A special emphasis for the 2019 workshop is the role of computing in drug discovery for cancer, with a special emphasis on the role of AI, machine learning, and deep learning approaches applied to drug discovery and precision medicine. The workshop will bring together the computational community and the cancer research community to explore applications of high-performance computing, analytics, predictive modeling, and large datasets in cancer research and clinical applications. The workshop is inherently inter-disciplinary, with the common interest in cancer and computation the unifying theme. As such, the workshop provides rich opportunities for attendees to learn about future directions, current applications and challenges and build collaborations. Maintaining a perspective of accelerating scientific insights and translation of insights to clinical application for improved patient outcomes, the workshop brings together many interests from across the technology, cancer research and clinical domains. Workshop Overview As a disease-specific workshop, the sustaining aim of the workshop is to provide a regular opportunity where individuals with expertise and interests across a variety of computational technologies and applications can convene, cross-educate, share progress, ideas and challenges, and seek opportunities for cross-interest collaboration and innovation. The participation in the workshop is inherently inter-disciplinary, with the common interest in cancer and computation serving as the unifying theme. As such, the workshop will provide rich opportunities for attendees to learn about diverse computational approaches and applications of HPC in cancer research and clinical applications. Maintaining a perspective of patient impact, with translation of ideas to clinical application, the workshop will include involvement of interests spanning from research to clinical application. Workshop Objectives * Bring together individuals employing computation in studying, diagnosing, treating, or preventing cancer * Discuss and bring forward with a special emphasis the opportunities and associated challenges that are shaping the way HPC and AI are used for cancer drug discovery * Attract and involve individuals with HPC and computational skills in technical areas that would contribute to the understanding, diagnostics, treatment and/or prevention of cancer * Provide an overview of current applications of computational approaches in several cancer research areas and clinical application * Educate attendees including students, grad students and those seeking to work across disciplines on the ways that computation is used in several areas of cancer research and clinical applications * Explore future opportunities where employing HPC will help advance cancer research and clinical communities * Computational and HPC approaches supporting cancer research and clinical applications * Exascale and machine learning in cancer applications * Special Workshop Focus ‘AI, Computing and Cancer Drug Discovery’ * Frontiers, collaboration and innovation Call for Papers * Program includes presentations from submitted abstracts as well as invited presenters * Call for paper details with key dates below * Submission: September 15, 2019 * Notification of Acceptance: October 1, 2019 * Workshop: November 17, 2019 Outcomes and Impact * Educate and raise awareness of the impact that HPC has in the cancer research and clinical domains * Build collaborative connections among attendees, addressing problems * Extend the community with adding to the growing email lists * Develop workshop proceedings to extend the visibility and impact of the workshop * Catalyze discussions on additional computing and cancer workshops in other conferences, further building the community nationally and internationally Organizing Committee * Sean Hanlon – NCI Center for Strategic Scientific Initiatives * Patricia Kovatch – Mount Sinai Icahn School of Medicine * Eric Stahlberg – Frederick National Laboratory for Cancer Research * Amanda Minnich – Lawrence Livermore National Laboratory

1 0

Re: [Candle-aesp] priorities for the AESP project
by Yoo, Hyunseung 01 Jul '19

01 Jul '19

I have a simple, master-slave model, parallel Mordred generator (https://github.com/hyoo/parallel_mordred_generator) and tested on a linux box. As I mentioned during the conference, there was a configuration issue; Theta wasn’t happy with conda-installed mpi4py, so I moved to mpi4py module and then the RDKit wasn’t compiled correctly. I will work with ALCF support team to resolve the issue and will try a couple of test runs to understand the throughput and will report back. Harry From: Candle-aesp <candle-aesp-bounces(a)lists.cels.anl.gov> on behalf of "Stevens, Rick L. via Candle-aesp" <candle-aesp(a)lists.cels.anl.gov> Reply-To: "Stevens, Rick L." <stevens(a)anl.gov> Date: Sunday, June 16, 2019 at 11:03 PM To: "Emily M. via Candle-aesp Dietrich" <candle-aesp(a)lists.cels.anl.gov> Subject: [Candle-aesp] priorities for the AESP project Hi, here are some things I’d like to make sure are moving forward. 1. Demonstration of our ability to compute drug descriptors for a large set of candidates in parallel. Ideally we would set up some runs that would compute mordred descriptors for each of the compound in the lists we have in this set of files. Is this something we can get running on Theta? 2. Prototype of our generative model that can be run in parallel to sweep through latent space and generate large stream of SMILEs examples. I’d like to get a version of our VAE models that can run in inference model and do some runs to generate 1M, 10M, 100M example compounds by sampling the latent space in some clever fashion. Ideally I’d like to take the set of generated compounds and search against those in the datasets in #1 to figure out how many are duplicates and how many are novel. For anything novel I would like to compute the Mordred descriptors and store them along with the smiles. Perhaps the right way to do this is to have the inference model directly call mordred so that what comes out of the model is SMILES + Descriptors .. and round robin joined to the RNAseq and SNP features for the Tumors (both CL and PDX) 3. Prototype of our single drug prediction model with UQ in regression mode on growth that is optimized for inference and that can take input from streams or a file. The idea here is that we we will have some number of these models that are running in parallel with #2 With UQ running we want to get out of the model both an estimate of the growth for the given drug/tumor combination but also an estimate of the error on that estimate. This error estimate will need to come from the calibration curves from the UQ. 4. A way to capture the prediction for each generated drug of all the cell lines and PDX models (think of this as a vector of predictions of order ~1K values and error) that we would want to associate with the drug candidate. 5. Some comparison of the throughput of running #1 coupled to #3 vs #2 coupled to #3. Perhaps we can iterate on this via email this week —rick

3 3