Abhi Datta
@datta_science
Followers
927
Following
2K
Media
57
Statuses
354
Associate Professor @jhubiostat geospatial, statistics, machine learning, AI, environmental health, global Health.
Baltimore, MD
Joined September 2014
So proud and excited to see my former PhD advisee and @jhubiostat alumnus @ArkajyotiSaha2 join UC Irvine Statistics Department as an Assistant Professor! Wishing you continued success.
I'm thrilled to share that I've started as an assistant professor in the Department of Statistics at UC Irvine's Donald Bren School of ICS (@UCIbrenICS). This would not have been possible without my mentors @daniela_witten, Jacob Bien, @datta_science, and @nilanjan10c!
0
0
17
New paper w/ former @jhubiostat student Brian Gilbert! We introduce visGP—a visibility-graph-based Gaussian process for spatial analysis in non-convex domains. It preserves Euclidean covariances where valid while incorporating complex domain geometry. https://t.co/QSEfvU2g1d
academic.oup.com
Abstract. We present a new method for constructing valid covariance functions of Gaussian processes for spatial analysis in irregular, non-convex domains s
0
0
19
I will be teaching a short course on geospatial machine learning at the International Biometric Society conference. #IBC2024ATL Sign up here. https://t.co/Zz7MV7EXEr
1
1
18
New paper! We extend Gaussian process regression to handle distribution-valued covariates. Provably optimal, and not needing the usual density estimation step in distribution regression. Check out this simpler, more direct approach! 📊✨ https://t.co/Ida62WtQDh
projecteuclid.org
In this manuscript, we study scalar-on-distribution regression; that is, instances where subject-specific distributions or densities are the covariates, related to a scalar outcome via a regression...
Excited to share new work with Bohao Tang, @yi_zhao1026, Brian Caffo. We generalize Gaussian Process regression to distribution-valued covariates, like high-frequency repeated measurements from sensors or outputs from probabilistic algorithms like MCMC. https://t.co/ESxaeh5qIQ
0
2
32
We first honor the eight recipients of the 2024 COPSS Emerging Leader Award! 👏 Congratulations to @datta_science, @Anru_Zhang, @baileyfosdick, @DanieleDurante2, Jennifer Bobb, Sandra Safo, Shu Yang, and Zheng Tracy Ke!
0
15
68
Congratulations to @elizabethTchin, @datta_science, Sandipan Pramanik, and @ScottZeger who were each part of teams who received an inaugural 2024 Data Science and AI Institute Demonstration Projects Award. Read more about their projects online: https://t.co/5xoTKrgW1w
0
2
16
Want to use neural networks for geospatial analysis, but not sure how to account for spatial correlation in the data? Check out the package geospaNN (geospatial neural networks) https://t.co/W0WtWllvrx The final version of the paper is also out now. https://t.co/hpYzetAHEX
🚀 Excited to share new paper led by @jhubiostat PhD student Wentao Zhang. We show the limitations of traditional neural networks for geospatial data and introduce a GNN that explicitly models spatial correlations. https://t.co/dA3r2JIswU Software: https://t.co/W0WtWlm3h5
0
7
26
📢 Attention All Graduate Students! 🌟 🎓 Attending #JSM2024? Don't miss out on the Graduate Student Mentoring Session organized by ASA Section on Statistics and Environment (ENVR)! https://t.co/CNc5QV1prs Join in-person (with lunch) or online. Sign-up: https://t.co/PuQWwSrzm1
0
3
11
🚀 New paper with former @jhubiostat student Claire Heffernan & @rdpeng @KirstenKoehler @drew_gentner @MistiLevyZamora Did the COVID-19 lockdowns affect air quality? We present a causal perspective using machine-learning based interrupted time series. https://t.co/vi1K094ghk
academic.oup.com
Abstract. When studying the impact of policy interventions or natural experiments on air pollution, such as new environmental policies or the opening or cl
0
3
9
My paper is now published in Biometrics: https://t.co/qpLGPWb2NO. Building off recent work, I develop a fast and exact way to obtain confidence intervals for the constant treatment effect when using the ANCOVA test statistic in a randomization inference framework
academic.oup.com
ABSTRACT. Randomization-based inference using the Fisher randomization test allows for the computation of Fisher-exact P-values, making it an attractive op
2
8
38
🚀 Excited to share new paper led by @jhubiostat PhD student Wentao Zhang. We show the limitations of traditional neural networks for geospatial data and introduce a GNN that explicitly models spatial correlations. https://t.co/dA3r2JIswU Software: https://t.co/W0WtWlm3h5
Very excited about this new work with PhD student Wentao Zhan on method and theory for neural networks for spatial data. We embed neural networks in traditional geospatial models to relax the linear mean assumption while still modeling the covariance using Gaussian Processes. 1/
0
11
47
Congratulations to Dr. Claire Heffernan, freshly minted PhD from @jhubiostat ! 🥳🎉 Witnessing the completion of a PhD is always a moment to cherish as an academic, celebrating both the achievement and the journey.🎓
0
1
48
Honored to be among the recipients of the 2024 COPSS Emerging Leader Award. https://t.co/JUnyvKqv0M
🎉 👏 Excited to announce our 2024 COPSS Emerging Leader Award (ELA) winners. The COPSS ELA recognizes early career statistical scientists who show evidence of and potential for leadership and who will help shape and strengthen the field.
8
1
48
Will be presenting at The International Environmetrics Society @TIES_isi webinar today and sharing my experience on using spatial machine-learning for environmental data. https://t.co/538W1X3mv1
0
0
14
Finally out in the Annals of Applied Statistics. https://t.co/ErnnZ83EOx
New paper forthcoming at Annals of Applied Statistics led by PhD student Claire Heffernan & with @rdpeng @KirstenKoehler @drew_gentner We show that regression-based calibration of low-cost sensor air-pollution data underestimates pollution peaks. 1/ https://t.co/JtG41bWLxj
1
0
22
Looking forward to visiting Yale Biostatistics @ysphbiostat today. Giving a seminar on model-free generalized Bayes for compositional data and application to calibration of verbal autopsies for improving national cause-specific mortality estimates. https://t.co/jww6acQwxs
0
0
19
Visiting @CornellDSDS today for a seminar. Looking forward to speaking about machine learning for spatially dependent data by incorporating data covariances into the algorithms. https://t.co/f4wNoySEMY
0
0
11
It was great to participate in the @JHU_India panel on AI for Climate and Health in India yesterday at the #555Penn building in DC. 🌍💡Insightful and diverse perspectives shared on the opportunities and challenges of AI in shaping climate and health research and policy. 🌿🤝
0
0
10
Important discussions at the #MITSAlliance summit in Nairobi on the potential of MITS (Minimally Invasive Tissue Sampling) for understanding cause-of-death at individual and population level. Shared our work on using MITS to calibrate verbal autopsies.🌍📊 @RTI_Intl #GlobalHealth
0
0
5