new!
📝 Mean Flows as Controllable Priors for Solving Inverse Problems
🏛️ NeurIPS Workshop on Beyond Next Token Prediction
✍️ Abdelaal, Du, Karimi
new!
📝 RecourseBench: A Modular Framework for Reproducible Algorithmic Recourse Evaluation
🏛️ arXiv preprint arXiv:2606.16113
✍️ Khotanlou, Ahmed, Tan, Abdelaal, Karimi
✨ TL;DR: RecourseBench provides a modular, reproducible, and interactive benchmark for evaluating algorithmic recourse methods
new!
📝 A Unified Perturbation Framework for Analyzing Leaderboard Stability and Manipulation
🏛️ NeurIPS 2026
✍️ Oyarhoseini, Lin, Karimi
✨ TL;DR: A perturbation framework for auditing and stress-testing the robustness of pairwise-comparison leaderboards
📝 Explainable AI is Causality in Disguise
🏛️ EurIPS Theory of XAI Workshop, 2025
✍️ Karimi
✨ TL;DR: Solving XAI requires us to make explicit our causal assumptions of the world
📝 Your Recourse, My Loss? Algorithmic Recourse under Shared Constraints
🏛️ ACM FAccT 2026
✍️ Khotanlou, Larson, Karimi
✨ TL;DR: Providing recourse to any one individual affects what recourse can be provided to other individuals
📝 Imagining and building wise machines: The centrality of AI metacognition
🏛️ Trends in Cognitive Sciences
(impact factor: 17.2)
✍️ Johnson, Karimi, Bengio, Chater, Gerstenberg, Larson, Levine, Mitchell, Rahwan, Schölkopf, Grossmann
✨ TL;DR: AI systems should integrate metacognitive strategies to enhance robustness, explainability, cooperation, and safety in complex real-world scenarios
📝 Temporal Convolutional Transformer for EEG Based Motor Imagery Decoding
🏛️ Nature Scientific Reports
(impact factor: 3.9)
✍️ Altaheri, Karray, Karimi
✨ TL;DR: TCFormer leverages CNNs, transformers, and temporal convolutions to decode EEG motor imagery, significantly outperforming existing BCI approaches
📝 On the Relationship Between Explanation and Prediction: A Causal View
🏛️ ICML 2023
(acceptance rate: 27.9%)
✍️ Karimi, Muandet, Kornblith, Schölkopf, Kim
✨ TL;DR: do explanations actually explain the predictions? yes, among other things....
📝 A Survey of Algorithmic Recourse: Contrastive Explanations & Consequential Recommendations
🏛️ ACM Computing Surveys 2022
(impact factor: 24.0)
✍️ Karimi, Barthe, Schölkopf, Valera
✨ TL;DR: definitions, formulations, solutions, and prospects for recourse research
📝 Key-Value Memory Networks for Directly Reading Documents
🏛️ EMNLP 2016
(acceptance rate: 24.3%)
✍️ Miller, Fisch, Dodge, Karimi, Bordes, Weston
✨ TL;DR: use key-value pairs to efficiently retrieve and store information for improved document comprehension
new!
📝 Mean Flows as Controllable Priors for Solving Inverse Problems
🏛️ NeurIPS Workshop on Beyond Next Token Prediction
✍️ Abdelaal, Du, Karimi
new!
📝 A Unified Perturbation Framework for Analyzing Leaderboard Stability and Manipulation
🏛️ NeurIPS 2026
✍️ Oyarhoseini, Lin, Karimi
✨ TL;DR: A perturbation framework for auditing and stress-testing the robustness of pairwise-comparison leaderboards
new!
📝 Beyond Explainable AI (XAI): An Overdue Paradigm Shift and Post-XAI Research Directions
🏛️ arXiv preprint arXiv:2602.24176
✍️ Afroogh, ..., Karimi, ...
✨ TL;DR: a cross-disciplinary research agenda for moving beyond current explainable AI paradigms toward post-XAI directions
new!
📝 How social context shapes value-related content in Large Language Model outputs
🏛️ COLM 2026
✍️ Ghorbansabagh, Karimi, Grossmann
📝 Explainable AI is Causality in Disguise
🏛️ EurIPS Theory of XAI Workshop, 2025
✍️ Karimi
✨ TL;DR: Solving XAI requires us to make explicit our causal assumptions of the world
📝 Prospector Heads: Generalized Feature Attribution for Large Models & Data
🏛️ ICML 2024
(acceptance rate: 27.5%)
✍️ Machiraju, Derry, Desai, Guha, Karimi, Zou, Altman, Ré, Mallick
📝 On the Relationship Between Explanation and Prediction: A Causal View
🏛️ ICML 2023
(acceptance rate: 27.9%)
✍️ Karimi, Muandet, Kornblith, Schölkopf, Kim
✨ TL;DR: do explanations actually explain the predictions? yes, among other things....
📝 On Data Manifolds Entailed by Structural Causal Models
🏛️ ICML 2023
(acceptance rate: 27.9%)
✍️ Dominguez-Olmedo, Karimi, Arvanitidis, Schölkopf
📝 Distance Correlation Autoencoder
🏛️ IJCNN 2018
(acceptance rate: 22.7%)
✍️ Wang, Karimi, Ghodsi
📝 SRP: Efficient class-aware embedding learning for large-scale data via supervised random projections
🏛️ arXiv
✍️ Karimi, Wong, Ghodsi
📝 Ensembles of Random Projections for Nonlinear Dimensionality Reduction
🏛️ CVIS 2017
(acceptance rate: 40.0%)
✍️ Karimi, Shafiee, Ghodsi, Wong
📝 Key-Value Memory Networks for Directly Reading Documents
🏛️ EMNLP 2016
(acceptance rate: 24.3%)
✍️ Miller, Fisch, Dodge, Karimi, Bordes, Weston
✨ TL;DR: use key-value pairs to efficiently retrieve and store information for improved document comprehension
📝 Spatio-temporal Saliency Detection using Abstracted Fully-Connected Graphical Models
🏛️ ICIP 2016
(acceptance rate: 40.0%)
✍️ Karimi, Shafiee, Scharfenberger, BenDaya, Haider, Talukdar, Clausi, Wong
new!
📝 RecourseBench: A Modular Framework for Reproducible Algorithmic Recourse Evaluation
🏛️ arXiv preprint arXiv:2606.16113
✍️ Khotanlou, Ahmed, Tan, Abdelaal, Karimi
✨ TL;DR: RecourseBench provides a modular, reproducible, and interactive benchmark for evaluating algorithmic recourse methods
📝 Your Recourse, My Loss? Algorithmic Recourse under Shared Constraints
🏛️ ACM FAccT 2026
✍️ Khotanlou, Larson, Karimi
✨ TL;DR: Providing recourse to any one individual affects what recourse can be provided to other individuals
📝 Causal Adversarial Perturbations for Individual Fairness and Robustness in Heterogeneous Data Spaces
🏛️ AAAI 2023
(acceptance rate: 23.7%)
✍️ Ehyaei, Mohammadi, Karimi, Samadi, Farnadi
📝 Robustness Implies Fairness in Causal Algorithmic Recourse
🏛️ ACM FAccT 2023
(acceptance rate: 24.6%)
✍️ Ehyaei, Karimi, Schölkopf, Maghsudi
✨ TL;DR: as the title says...
📝 On the Robustness of Causal Algorithmic Recourse
🏛️ ICML 2022
(acceptance rate: 21.9%)
✍️ Dominguez-Olmedo, Karimi, Schölkopf
✨ TL;DR: robustness of recourse is separate from robustness of prediction
📝 On the Fairness of Causal Algorithmic Recourse
🏛️ AAAI 2022
(acceptance rate: 15.0%)
✍️ von Kügelgen, Karimi, Bhatt, Valera, Weller, Schölkopf
✨ TL;DR: fairness of recourse is separate from fairness of prediction
📝 A Survey of Algorithmic Recourse: Contrastive Explanations & Consequential Recommendations
🏛️ ACM Computing Surveys 2022
(impact factor: 24.0)
✍️ Karimi, Barthe, Schölkopf, Valera
✨ TL;DR: definitions, formulations, solutions, and prospects for recourse research
📝 Algorithmic Recourse under Imperfect Causal Knowledge: a Probabilistic Approach
🏛️ NeurIPS 2020
(acceptance rate: 20.1%)
✍️ Karimi‡, von Kügelgen‡, Schölkopf, Valera
✨ TL;DR: causal assumptions are rarely available; instead generate recourse probabilistically
📝 Algorithmic Recourse: from Counterfactual Explanations to Interventions
🏛️ ACM FAccT 2021
(acceptance rate: 25.0%)
✍️ Karimi, Schölkopf, Valera
✨ TL;DR: real-world causal assumptions are needed for counterfactual explanations & recourse
📝 Scaling Guarantees for Nearest Counterfactual Explanations
🏛️ ACM AIES 2021
(acceptance rate: 38.0%)
✍️ Mohammadi, Karimi, Barthe, Valera
✨ TL;DR: generate optimal counterfactual explanations using mixed-integer linear programs
📝 Model-Agnostic Counterfactual Explanations for Consequential Decisions
🏛️ AISTATS 2019
(acceptance rate: 32.4%)
✍️ Karimi, Barthe, Valera
✨ TL;DR: generate optimal counterfactual explanations using satisfiability solvers
new!
📝 Metacognition Should Be the Scientific Framework for Bounded and Effective Self-Governance in Generative AI
🏛️ arXiv preprint arXiv:2605.23981
✍️ Ji, Grossmann, Karimi
✨ TL;DR: Metacognition offers a framework for bounded and effective self-governance in generative AI systems
📝 Imagining and building wise machines: The centrality of AI metacognition
🏛️ Trends in Cognitive Sciences
(impact factor: 17.2)
✍️ Johnson, Karimi, Bengio, Chater, Gerstenberg, Larson, Levine, Mitchell, Rahwan, Schölkopf, Grossmann
✨ TL;DR: AI systems should integrate metacognitive strategies to enhance robustness, explainability, cooperation, and safety in complex real-world scenarios
📝 Temporal Convolutional Transformer for EEG Based Motor Imagery Decoding
🏛️ Nature Scientific Reports
(impact factor: 3.9)
✍️ Altaheri, Karray, Karimi
✨ TL;DR: TCFormer leverages CNNs, transformers, and temporal convolutions to decode EEG motor imagery, significantly outperforming existing BCI approaches
📝 FEELS: a Full-Spectrum Enhanced Emotion Learning System for Assisting People with Autism Spectrum Disorder
🏛️ CVIS 2018
(acceptance rate: 40.0%)
✍️ Karimi, Boroomand, Pfisterer, Wong
📝 Synthesizing Deep Neural Network Architectures using Biological Synaptic Strength Distributions
🏛️ CCN 2017
(acceptance rate: 40.0%)
✍️ Karimi, Shafiee, Ghodsi, Wong
📝 Discovery Radiomics via a Mixture Sequencers for Multi-Parameter MRI
🏛️ ICIAR 2017
(acceptance rate: 40.0%)
✍️ Karimi, Chung, Shafiee, Khalvati, Haider, Ghodsi, Wong
📝 Automated detection and cell density assessment of keratocytes in the human corneal stroma from ultrahigh resolution optical coherence tomograms
🏛️ Biomedical Optics Express
(impact factor: 2.9)
✍️ Karimi, Wong, Bizheva