Keywords: Interpretability Science Discovery
ML has shown great promise in modeling and predicting complex phenomenon in many scientificdisciples such as predicting cardiovascular risk factors from retinal images, understanding howelectrons behave at the atomic level [3], identifying patterns of weather and climate phenomena, etc. Further, models are able to learn directly (and better) from raw data as opposed to human selected features. The ability to interpret the model and find significant predictors couldprovide new scientific insights.Traditionally, the scientific discovery process has been based on careful observations of nat-ural phenomenon, followed by systematic human analysis (of hypothesis generation and ex-perimental validation). ML interpretability has the potential to bring a radically different yetprincipled approach. While general interpretability relies on ‘human parsing’ (common sense),scientific domains have semi-structured and highly structured bases for interpretation. Thus,despite differences in data modalities and domains, be it brain sciences, the behavioral sciences,or material sciences, there is a need for a common set of tools that address a similar flavor of problem, one of interpretability or fitting models to a known structure. This workshop aims to bring together members from the ML and physical sciences communities to introduce exciting problems to the broader community, and stimulate the productionof new approaches towards solving open scientific problems.
Fri 6:50 a.m. - 7:00 a.m.
|
Welcome Note
(
Talk
)
|
Subhashini Venugopalan 🔗 |
Fri 7:00 a.m. - 7:50 a.m.
|
Invited Talk - Eun-Ah Kim
(
Talk followed by Q&A
)
|
Eun-Ah Kim · Subhashini Venugopalan 🔗 |
Fri 7:50 a.m. - 8:30 a.m.
|
Invited Talk - Barbara Engelhardt
(
Talk
)
|
Barbara Engelhardt · Subhashini Venugopalan 🔗 |
Fri 8:30 a.m. - 8:40 a.m.
|
Q&A - Barbara Engelhardt
(
Q&A
)
|
Barbara Engelhardt · Subhashini Venugopalan 🔗 |
Fri 8:40 a.m. - 9:20 a.m.
|
Invited Talk - Sendhil Mullainathan, Machine Learning for Scientific Discovery (recorded)
(
Talk
)
link »
SlidesLive Video » |
Sendhil Mullainathan · Subhashini Venugopalan 🔗 |
Fri 9:20 a.m. - 9:30 a.m.
|
Q&A - Sendhil Mullainathan
(
Q&A
)
|
Sendhil Mullainathan · Subhashini Venugopalan 🔗 |
Fri 9:30 a.m. - 10:10 a.m.
|
Invited Talk - Arun Narayanaswamy, ML Driven Scientific Discovery (recorded)
(
Talk
)
link »
SlidesLive Video » |
Arunachalam Narayanaswamy · Subhashini Venugopalan 🔗 |
Fri 10:10 a.m. - 10:20 a.m.
|
Q&A - Arun Narayanaswamy
(
Q&A
)
|
Subhashini Venugopalan 🔗 |
Fri 10:20 a.m. - 10:50 a.m.
|
Break - Poster videos (recorded)
link »
Break + Watch some poster presentation videos Watch poster videos here |
Subhashini Venugopalan 🔗 |
Fri 10:50 a.m. - 11:40 a.m.
|
Invited Talk - Katie Bouman
(
Talk followed by Q&A
)
|
Katherine Bouman · Subhashini Venugopalan 🔗 |
Fri 11:40 a.m. - 11:50 a.m.
|
Panel Discussion - setup
|
Subhashini Venugopalan 🔗 |
Fri 11:50 a.m. - 1:05 p.m.
|
Panel Discussion
|
Subhashini Venugopalan · Eun-Ah Kim · Barbara Engelhardt · Sendhil Mullainathan · Arunachalam Narayanaswamy · Katherine Bouman 🔗 |
Fri 1:10 p.m. - 1:35 p.m.
|
Poster Session (recorded)
(
Poster Videos
)
link »
Watch poster videos here Accepted papers pdf and videos are also here: https://sites.google.com/corp/view/mli4sd-icml2020/program |
Subhashini Venugopalan 🔗 |
Fri 1:35 p.m. - 1:45 p.m.
|
Parallel poster session info, Closing remarks
(
Q&A
)
Closing remarks + Logistics for parallel poster sessions and break out. |
Subhashini Venugopalan 🔗 |
Fri 1:45 p.m. - 2:45 p.m.
|
Parallel poster session
(
Poster & networking session
)
link »
Poster and Networking session How does the || poster session work: Each poster has a zoom room where one of the authors will be present to answer questions. Click on the poster to see the zoom link information in the abstract. Use the gather.town link below to network, and take your poster sessions outside of zoom after your session ends. [ protected link dropped ] |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#05 Actionable Attribution Maps for Scientific Machine Learning
(
Poster
)
link »
Actionable Attribution Maps for Scientific Machine Learning Shusen Liu (Lawrence Livermore National Laboratory)*; Bhavya Kailkhura (Lawrence Livermore National Laboratory); Jize Zhang (Lawrence Livermore National Laboratory); Anna Hiszpanski (Lawrence Livermore National Laboratory); Emily Robertson (Lawrence Livermore National Laboratory); Donald Loveland (Lawrence Livermore National Laboratory); T. Yong-Jin Han (LLNL) Paper: https://arxiv.org/abs/2006.16533 Video: https://youtu.be/IMfWZ82Dcow Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#06 Unpacking Chemical Reaction Prediction Models Using Integrated Gradients
(
Poster
)
link »
Unpacking Chemical Reaction Prediction Models Using Integrated Gradients William McCorkindale (University of Cambridge); David P Kovacs (University of Cambridge)*; Alpha Lee (University of Cambridge) Paper: https://www.dropbox.com/s/jrsfvtbwnxqxwgm/ICML2020_IGs-4.pdf?dl=0 Video: https://drive.google.com/file/d/1j011oIdflLa_Hry1ws-ZasdA5Y0A-STt/view Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#07 Inverse Problems, Deep Learning, and Symmetry Breaking
(
Poster
)
link »
Inverse Problems, Deep Learning, and Symmetry Breaking Kshitij Tayal (University of Minnesota)*; Chieh-Hsin Lai (University of Minnesota, Twin Cities); Raunak Manekar (University of Minnesota); Vipin Kumar (University of Minnesota); Ju Sun (University of Minnesota) Paper: https://sunju.org/docs/ICML20-WS-DL4INV.pdf Video: https://drive.google.com/drive/folders/1RT7xQ6QAvUhrb7EQxytkvOuNXj1Zdth_?usp=sharing Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#08 (Re)Discovering Protein Structure and Function Through Language Modeling
(
Poster
)
link »
(Re)Discovering Protein Structure and Function Through Language Modeling Jesse Vig (Salesforce)*; Ali Madani (Salesforce Research); Lav Varshney (UIUC: ECE); Nazneen Fatema Rajani (Salesforce Research) Paper: https://storage.googleapis.com/sfr-jvig-data-research/rediscovering_protein.pdf Video: https://vimeo.com/434882244 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#10 Explaining Chemical Toxicity Using Missing Features
(
Poster
)
link »
Explaining Chemical Toxicity Using Missing Features Kar Wai Lim (IBM Singapore)*; Bhanushee Sharma (Rensselaer Polytechnic Institute); Vijil Chenthamarakshan (IBM AI Research); Payel Das (IBM Research); Jonathan S. Dordick (Rensselaer Polytechnic Institute) Paper: https://github.com/karwailim/Explaining-Chemical-Toxicity-Using-Missing-Features/raw/master/MLI4SDW2020_Explaining_Chemical_Toxicity_Using_Missing_Features.pdf |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#11 Modeling Brain Microarchitecture with Deep Representation Learning
(
Poster
)
link »
Modeling Brain Microarchitecture with Deep Representation Learning Aishwarya H. Balwani (Georgia Institute of Technology)*; Eva Dyer (Georgia Tech) Paper: https://aishwaryahb.github.io/docs/papers/Balwani_ICML_Interpretability_Workshop_2020.pdf Video: https://www.dropbox.com/s/ra3o19eu4xdyywj/Modeling%20Brain%20Microarchitecture%20with%20Deep%20Representation%20Learning.mp4?dl=0 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#12 Feature Extraction on Synthetic Black Hole Images
(
Poster
)
link »
Feature Extraction on Synthetic Black Hole Images Joshua Yao-Yu Lin (Physics department, University of Illinois at Urbana-Champaign)*; George Wong (University of Illinois at Urbana-Champaign); Ben Prather (University of Illinois at Urbana-Champaign); Charles Gammie ( Paper: https://arxiv.org/pdf/2007.00794.pdf Video: https://www.youtube.com/watch?v=D4QNmjXtp3Q Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#13 Learning about learning by many-body systems
(
Poster
)
link »
Learning about learning by many-body systems Weishun Zhong (Massachusetts Institute of Technology)*; Jacob Gold (Massachusetts Institute of Technology); Sarah Marzen (Massachusetts Institute of Technology; Claremont Colleges); Jeremy L England (GlaxoSmithKline); Nicole Yunger Halpern (Harvard University; Massachusetts Institute of Technology ) Paper: https://github.com/smarzen/Statistical-Physics/blob/master/many_body_learning_ICML_workshop_July%201st_2020.pdf.pdf Video: https://www.youtube.com/watch?v=X8MhJw-bhws&feature=youtu.be Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 1:45 p.m. - 2:15 p.m.
|
[Session 1] P#14 DisCont: Self-Supervised Visual Attribute Disentanglement using Context Vectors
(
Poster
)
link »
DisCont: Self-Supervised Visual Attribute Disentanglement using Context Vectors Sarthak Bhagat (IIIT-Delhi)*; Vishaal Udandarao (IIIT Delhi); Shagun Uppal (IIIT-Delhi) Paper: https://arxiv.org/abs/2006.05895 Video: https://youtu.be/2gyLEPnTi1M Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#15 Look at the Loss: Towards Robust Detection of False Positive Feature Interactions Learned by Neural Networks on Genomic Data
(
Poster
)
link »
Look at the Loss: Towards Robust Detection of False Positive Feature Interactions Learned by Neural Networks on Genomic Data Mara R Finkelstein (Stanford University); Avanti Shrikumar (Stanford University)*; Anshul Kundaje (Stanford University) Paper: https://github.com/kundajelab/feature_interactions/blob/master/Look_At_The_Loss_MLiSD_ICML_Workshop.pdf Video: https://drive.google.com/drive/folders/127QlmcQZpfCNWmhtOuA79yRCpqOtQnWV?usp=sharing Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#16 Attribution Methods Reveal Flaws in Fingerprint-Based Virtual Screening
(
Poster
)
link »
Attribution Methods Reveal Flaws in Fingerprint-Based Virtual Screening Vikram Sundar (Google); Lucy Colwell (Google)* Paper: https://arxiv.org/pdf/2007.01436.pdf Video: https://www.youtube.com/watch?v=2Vv_e53DuFs Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#17 In-Distribution Interpretability for Challenging Modalities
(
Poster
)
link »
In-Distribution Interpretability for Challenging Modalities Cosmas Heiss (TU Berlin)*; Ron Levie (TU Berlin); Cinjon Resnick (NYU); Gitta Kutyniok (Technische Universität Berlin); Joan Bruna (Courant Institute of Mathematical Sciences, NYU, USA) Paper: https://arxiv.org/abs/2007.00758 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#18 Extracting Interpretable Physical Parameters from Spatiotemporal Systems using Unsupervised Learning
(
Poster
)
link »
Extracting Interpretable Physical Parameters from Spatiotemporal Systems using Unsupervised Learning Peter Y Lu (MIT)*; Samuel Kim (MIT); Marin Soljacic (MIT) Paper: https://www.dropbox.com/s/z6xlp5lcktd4kss/ICML2020workshop_pde_vae_final.pdf Video: https://youtu.be/4tjD6_7Eclg Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#20 Deep Interpretability for GWAS
(
Poster
)
link »
Deep Interpretability for GWAS Deepak Sharma (MILA, McGill University)*; Audrey Durand (Université Laval); Marc-Andre Legault (Université de Montréal); Louis-Philippe Lemieux (Montreal Heart Institute Paper: https://arxiv.org/abs/2007.01516 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#24 Interpreting Stellar Spectra with Unsupervised Domain Adaptation
(
Poster
)
link »
Interpreting Stellar Spectra with Unsupervised Domain Adaptation Sébastien Fabbro (NRC Herzberg)*; Kwang Moo Yi (University of Victoria); Teaghan O'Briain (University of Victoria); Kim Venn (University of Victoria); Yuan-Sen Ting (Institute for Advanced Study Princeton); Spencer Bialek (University of Victoria) Paper: https://github.com/teaghan/Cycle_SN/raw/master/docs/ICMLW2020_StellarSpectra.pdf Video: https://github.com/teaghan/Cycle_SN/raw/master/docs/IMCLW2020_StellarSpectra_Video.mp4 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#27 End to End learning for Phase Retrieval
(
Poster
)
link »
End to End learning for Phase Retrieval Raunak Manekar (University of Minnesota); Kshitij Tayal (University of Minnesota)*; Vipin Kumar (University of Minnesota); Ju Sun (University of Minnesota) Paper: https://sunju.org/docs/ICML20-WS-DL4FPR.pdf Video: https://drive.google.com/drive/folders/10To09kLk0PP-M6kCqe5T4sJzxlnuHfz1?usp=sharing Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#28 Learning Cell State Representations From Barcoded Gene-Expression Trajectories
(
Poster
)
link »
Learning Cell State Representations From Barcoded Gene-Expression Trajectories Yu Wu (Princeton University)*; Le Cong (Stanford University); Mengdi Wang (Princeton University/DeepMind) Paper: https://www.dropbox.com/s/4hs9nohcy5449gw/cell_embedding_icml_2020.pdf?dl=0 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
Fri 2:15 p.m. - 2:45 p.m.
|
[Session 2] P#29 Unsupervised Attention-Guided Atom-Mapping
(
Poster
)
link »
Unsupervised Attention-Guided Atom-Mapping Philippe PS Schwaller (IBM Research Europe / University of Bern)*; Benjamin Hoover (IBM Research); Jean-Louis Reymond (University of Bern); Hendrik Strobelt (IBM Research); Teodoro Laino (IBM Research Europe) Paper: http://rxnmapper.org/demo/atom_mapping_ICMLWorkshop.pdf Video: https://vimeo.com/434757113 Please join poster Q&A zoom via link below. Please DO NOT SHARE zoom link externally. |
🔗 |
-
|
Networking Town
(
Networking
)
link »
|
🔗 |