ATLAS: Learning to Recommend Across Unseen Domains
What if a recommender trained only on movies and music could walk into a catalogue it has never seen, groceries or video games, and still know what to suggest?
RESEARCH
Published Work
Causal inference, counterfactual reasoning, and the question of why.
Papers
GENESIS: Towards Explainable Causal Discovery
If an algorithm hands you a map of cause and effect, should it not also be able to justify every single arrow it drew?
RecRec: Recursive Refinement for Sequential Recommendation
What if a recommender never judged you in a single glance, but kept revisiting its impression of you, refining it pass after pass, anchored to what you actually did?
KANITE: Kolmogorov-Arnold Networks for ITE Estimation
Can a new kind of neural network teach us to predict exactly how any one person would respond to a treatment before it is ever given?
I See, Therefore I Do: Estimating Causal Effects for Image Treatments
What if a photograph itself is the treatment and we could measure exactly how much it changed your decision?
From What to Why: Thought-Space Recommendation with Small Language Models
What if a recommendation system could understand not what you clicked but why you wanted it in the first place?
PlotTwist: A Creative Plot Generation Framework with Small Language Models
What if a compact model could write plots that move you, not by knowing more, but by first learning the difference between a story that grips and one that falls flat?
Estimation of Individual Causal Effects in Network Setup for Multiple Treatments
Does understanding the network effect between humans help us analyse cause and effect across multiple real-world scenarios?