February, 2024
New paper on calibrating graph neural networks to be presented at #ICLR2024 – A stochastic centering approach for uncertainty estimation and calibration. [Paper]
Three papers on AI safety and OOD generalization accepted for publication at #ICASSP2024 – (i) Is CLIP prior useful for obtaining generalizable visual relationship prediction? (ii) Balancing anomaly detection and OOD generalization using model anchoring and (iii) how do we estimate uncertainties in edge-predicting GNNs?
September, 2023
Codes for CREPE to learn CLIP-based representations for visual predicate estimation released! [Codes]
Our paper on diffusion models for limited-angle CT reconstruction to be presented as a poster at #ICCV2023. [Paper] [Website]
Oral talk at the #ICCV2023 workshop on Visual Inductive Priors for Data-Efficient Deep Learning – A tuning-free augmentation policy for data-efficient and robust object detection [Paper]
Our paper on medical OOD detection to be presented at the #ICCV2023 workshop on UQ for computer vision. [Paper] [Codes]
August, 2023
Two patents granted – (i) Universal image representation based on a multimodal graph and (ii) Systems and methods for time series analysis using attention models. Yayyy!!
New paper in #NatureMachineIntelligence presents the MedPerf project, which provides the tools and infrastructure to distribute AI models to healthcare facilities [Paper]
OODmedic is out now! It can be used to design medical OOD detectors to reject semantic and modality shifts! [Website]
July, 2023
Attending #ICML2023 in Hawaii this week!
Our paper on single-shot adaptation via target-aware augmentations to be presented as a poster at #ICML2023 [Paper] [Website]
Two posters accepted at the #ICML2023 Data-centric Machine Learning workshop on failure characterization [Paper] and graph uncertainty quantification [Paper]
Our recent work on GAN adaptation via inversion (AdvIn) to be presented at the #ICML2023 workshop on Challenges in Deployable AI [Paper]
Presented our paper (oral) on augmentation design for outlier-free OOD detector training at #MIDL2023 [Paper][Website]
June, 2023
Attending #CVPR2023 in Vancouver this week!
#CVPR2023 poster on auditing GAN models to perform attribute-level comparison of two or more StyleGANs in an unsupervised fashion [Paper]
Codes for Delta-UQ uncertainty estimator released!!! If you want to quickly integrate epistemic UQ into your deep model, check this out.
Attending #ICASSP2023 in Rhodes this week!
Two papers accepted for presentations at the #ICASSP2023 conference – (i) a closer look at scoring function design for generalization gap predictors [Paper] and (ii) generative augmentations for single-shot domain adaptation [Paper]
May, 2023
Attending #ICLR2023 in Kigali this week!
#ICLR2023 spotlight paper on adapting pre-trained representations to ensure generalization and safety [Paper]
January, 2023
New #IEEE Access article on the utility of simple deep subspace alignment in challenging domain adaptation settings [Paper]
Attending #WACV2023 in Hawaii this week! Looking forward to meeting my collaborators and friends
Two new papers to appear at #WACV2023: (i) Improving generalization of meta learners via contrastively-trained knowledge graph bridges [Paper] and (ii) Diversity or Adversity? What is more critical for domain generalization? [Paper]
December, 2022
Two new papers to appear at #ACML2022: (i) AMP uses a neural network anchoring based uncertainty estimates for prediction calibration [Paper] and (ii) Fully test-time adaptation meets domain alignment [Paper]
Presenting two papers at #NeurIPS2022: (i) Single-model uncertainty estimation using stochastic data centering (Spotlight) [Paper] and Analyzing data-centric properties for contrastive learning on graphs [Paper]
jjthiagarajan.com 