Karthika Mohan
@Carthica
Followers
1K
Following
202
Statuses
81
Asst. Professor (Computer Science), Oregon State University Postdoc UC Berkeley, PhD UCLA (J Pearl) Artificial Intelligence, Causal Inference, Graphical Models
Corvallis, Oregon
Joined October 2010
Honored to speak on causality at #IBC2024 in Atlanta! 🎉 Humbled to hear my work on missing data has helped improve outcomes for children. 🙏✨ Huge shout-out to the researchers at Murdoch Children's Research Institute! 👏🌟 @yudapearl @_MargaritaMB @JiaxinZhang96 @ghazalehdashti
0
7
28
Challenging the IID assumption is key to advancing AI/ML, as real-world data often violates IID. We are excited to present our paper that explores this frontier.
Do Finetti: On Causal Effects for Exchangeable Data. #NeurIPS2024 Oral Session 5B: Fri 13 Dec 10 a.m. — 11 a.m. PST Poster: West Ballroom A-D #5000 Fri 13 Dec 11 a.m. — 2 p.m. PST joint work with @zcccucla @Carthica @fhuszar @bschoelkopf
0
1
16
@devendratweetin @EricREaton Thank you for the shoutout! Excited to give my talk this February at Continual Causality, AAAI-25. Looking forward to it!
0
0
2
We invite submissions of extended abstracts, up to 2 pages in length, to the UAI Causal Inference Workshop.
Excited to announce the 9th #causality workshop at @UncertaintyInAI with @raziehnabi @Carthica @CalebMiles16 @d_malinsky 🎉 Do you have interesting work in #causality #causalInference #causalDiscovery #CRL etc? Consider submitting an abstract by May 31st:
0
1
9
@yudapearl This is truly saddening. I recall how much you enjoyed his book and recommended it to me.
1
0
3
I discussed the application of Operations Research methods to the challenge of Causal Discovery with Imperfect Data at the Computing Community Consortium (CCC) AI/OR Workshop in Washington DC. Grateful for the opportunity and huge thanks to the fantastic organizers!
In the third technical presentation of the AI / OR workshop, Karthika Mohan talks about how to use integer programming for causal inference over incomplete data #orms #monarchsofmip
0
4
16
0
0
1
Enjoying every page of this book! Delighted by the dedicated chapter on causality, thrilled with the missing data section, and grateful for the shoutout to my work with @yudapearl . A must-read for AI students.
0
2
23
RT @yudapearl: Remember PO's slogan "causal inference is a missing-data problem"? Well, here @Carthica shows the opposite: "missing-data is…
0
4
0
Congratulations to Dr. Chi Zhang (@zcccucla) on defending her PhD thesis! Her work on interference pushes research boundaries and highlights the perils of blindly assuming IID. Well done, Dr. Zhang! It's been a delightful journey collaborating with you. 🥳🎉🍾 @yudapearl @oacarah
1
2
28
@AngeloDalli @yudapearl O and Ro are parents of O*. You can model the scenario by adding two edges, one between U & O and the other between U & Ro.
1
0
1
@raziehnabi @analisereal @noah_greifer @stephenjwild @r0ntu Agree! However, recoverability is query dependent. In X-->Y<-->Rx, P(X) is recoverable (despite Y being a collider between X and Rx) but P(Y,X) is not. Reference Mohan & Pearl, Neurips-14,
0
0
3
@yudapearl do you understand the analogy to a right angled triangle that Don Rubin makes in his interview: on pages 88 and 89 ??? I am bewildered.
0
1
5
0
0
1
Is it safe to assume IID when you know that data are not IID? What conditions contribute to bias when estimating causal effects? How can we remove bias due to interference? Answers to these and more are discussed in our paper!
0
42
220
@KirkDBorne Also, there are algorithms that are faster and more efficient (compared to EM) for parameter learning under missingness. See
0
0
1