Assistant Professor, Associate Director, VinUni-Illinois Smart Health Center, VinUniversity. At a Vietnamese institution since 2022.
Also published as: Đoàn Đăng Khoa, Khoa Doan, Khoa Dang Doan, Dang Khoa Doan
20 accepted papers at the ten venues since 2022: 16 in the official lists and 4 known only from personal pages, shown as +4 and tagged "unofficial" below. 11 accepted before 2026, 9 from 2026. ICLR: 14 submitted, 3 accepted, 11 rejected. Rejected and on record at other venues (public rejections and papers placed in Findings): NeurIPS 2, CVPR 1, ACL 1.
Links
- Homepage: khoadoan.me
- OpenReview profile (~Khoa_Doan1): openreview.net/profile?id=~Khoa_Doan1
- Publications: mail-research.com/publications
- Website: mail-research.com
- DBLP: dblp.org/pid/238/4276
- Google Scholar: scholar.google.com/citations?user=Zz2hMgcAAAAJ
Co-authors
- Thanh-Binh Nguyen (HCMUS): 4 papers
- An Thai Le (VinUni): 2 papers
- Binh T. Nguyen (VinUni): 2 papers
- An Mai (HCMIU): 1 paper
- Danh Le-Phuoc (VinUni): 1 paper
- Minh Do (VinUni): 1 paper
- Ngo Van Linh (HUST): 1 paper
- Tuan Dam (HUST): 1 paper
People on this site who share papers with this professor, accepted, rejected or workshop, with their institution today and the number of shared papers; see Researcher collaboration for the whole graph.
Google Scholar
A snapshot of the Google Scholar profile taken on 2026-10-02; it is not updated. Google Scholar counts every publication, not only the ten conferences.
| All | Since 2021 | |
|---|---|---|
| Citations | 1969 | 1924 |
| h-index | 15 | 15 |
| i10-index | 24 | 23 |
Submitted and accepted per year
Accepted papers per year by where the authors are
Each column is one year's accepted papers; grey papers are known only from announcements or personal pages.
Co-authors and topics
- 5.1 co-authors per paper on average, accepted or not (fewest on an accepted paper 2, most on any submission 22), 102 different co-authors over 35 papers.
- 3.2 foreign co-authors per paper on average: co-authors whose affiliation on the paper is outside Vietnam.
- 62 different co-authors are at another university or institution than VinUni (46 abroad, 16 in Vietnam).
- The last two figures use the 20 of 35 papers whose list prints an affiliation for each author.
- 7 topics in 3 areas, from the conference's own label on 8 of 35 papers: Deep Learning (3), Social Aspects (3), Applications (2).
| Topic area | Topic | Papers |
|---|---|---|
| Social Aspects | Trustworthy Machine Learning | 2 |
| Applications | Chemistry and Drug Discovery | 1 |
| Applications | Robotics | 1 |
| Deep Learning | Foundation Models | 1 |
| Deep Learning | Large Language Models | 1 |
| Deep Learning | Robustness | 1 |
| Social Aspects | Accountability, Transparency and Interpretability | 1 |
Accepted papers (20)
-
ICLR 2026
FACET: A Fragment-Aware Conformer Ensemble Transformer
-
NeurIPS 2026unofficial
Decoding the Critique Mechanism in Large Reasoning Models
Unofficial: name, and listed on their own page; in the NeurIPS 2026 list, not yet confirmable on OpenReview.
-
NeurIPS 2026unofficial
On the Primacy Bias in RLVR Training
Unofficial: name, and listed on their own page; in the NeurIPS 2026 list, not yet confirmable on OpenReview.
-
NeurIPS 2026unofficial
Rethinking Molecular Graph Backdoors under Chemistry-aware Admission
Unofficial: name, and listed on their own page; in the NeurIPS 2026 list, not yet confirmable on OpenReview.
-
NeurIPS 2026unofficial
The Reasoning Boundary Paradox: How Reinforcement Learning Constrains Language Models
Unofficial: name, and listed on their own page; in the NeurIPS 2026 list, not yet confirmable on OpenReview.
-
ICML 2026
FOCA: Future-Oriented Conditioning for Data-Efficient Vision-Language-Action Adaptation
-
ICML 2026
TokenRatio: Principled Token-Level Preference Optimization via Ratio Matching
-
CVPR 2026
HFedATM: Hierarchical Federated Domain Generalization via Optimal Transport and Regularized Mean Aggregation
-
AAAI 2026
Clean-Label Physical Backdoor Attacks with Data Distillation
-
ICLR 2025
Wicked Oddities: Selectively Poisoning for Effective Clean-Label Backdoor Attacks
-
NeurIPS 2025
How Many Tokens Do 3D Point Cloud Transformer Architectures Really Need?
-
NeurIPS 2025
Mitigating Reward Over-optimization in Direct Alignment Algorithms with Importance Sampling
-
NeurIPS 2025
Unveiling Concept Attribution in Diffusion Models
-
ICML 2025oral
LLM-SRBench: A New Benchmark for Scientific Equation Discovery with Large Language Models
-
ICLR 2024
Understanding the Robustness of Randomized Feature Defense Against Query-Based Adversarial Attacks
-
ECCV 2024
Data Poisoning Quantization Backdoor Attack
-
ECCV 2024oral
Flatness-aware Sequential Learning Generates Resilient Backdoors
-
NeurIPS 2023
IBA: Towards Irreversible Backdoor Attacks in Federated Learning
-
AAAI 2023
Defending Backdoor Attacks on Vision Transformer via Patch Processing
-
NeurIPS 2022
Marksman Backdoor: Backdoor Attacks with Arbitrary Target Class
Rejected (15)
Submissions that were not accepted.
-
ICLR 2026, rejected
The Reasoning Boundary Paradox: How Reinforcement Learning Constrains Language Models
The same title was accepted at NeurIPS 2026.
-
CVPR 2026 · Findings, rejected
Onboarding Without Forgetting: Hypernetwork Personalization with Data-Free Replay for Personalized Federated Learning
-
ICLR 2025, rejected
Are you SURE? Enhancing Multimodal Pretraining with Missing Modalities through Uncertainty Estimation
-
ICLR 2025, rejected
Forget but Recall: Incremental Latent Rectification in Continual Learning
-
ICLR 2025, rejected
Mitigating Reward Over-optimization in Direct Alignment Algorithms with Adaptive Importance Sampling
The same title was accepted at NeurIPS 2025.
-
ICLR 2025, rejected
Overcoming Catastrophic Forgetting in Federated Class-Incremental Learning via Federated Global Twin Generator
-
ICLR 2025, rejected
Sparse Watermarking in LLMs with Enhanced Text Quality
-
ICLR 2025, rejected
Synthesizing Physical Backdoor Datasets: An Automated Framework Leveraging Deep Generative Models
-
ICLR 2025, rejected
Unveiling Concept Attribution in Diffusion Models
The same title was accepted at NeurIPS 2025.
-
NeurIPS 2025, rejected
TriggerCraft: A Framework for Enabling Scalable Physical Backdoor Dataset Generation with Generative Models
-
NeurIPS 2025, rejected
Weak-to-strong Generalization via Formative Learning from Student Demonstrations & Teacher Evaluation
-
ICLR 2024, rejected
Fooling the Textual Fooler via Randomizing Latent Representations
-
ICLR 2024, rejected
Forget-Me-Not: Making Backdoor Hard to be Forgotten in Fine-tuning
-
ACL 2024, rejectedunofficial
Fooling the Textual Fooler via Randomizing Latent Representations
Unofficial: listed on their own page; not in the official list yet.
-
ICLR 2022, rejected
Gradient Boosting Neural Networks: GrowNet
Not main track (1)
Datasets & Benchmarks, journal and other tracks, which the site does not count.
-
ICML 2026
Retrospective Feature Estimation for Continual Learning
Not counted: journal track.
Before joining a Vietnamese institution (1)
Papers from before the year this person's post in Vietnam began, which the site does not count.
-
NeurIPS 2021
Backdoor Attack with Imperceptible Input and Latent Modification
Not counted: before 2022, when they joined a Vietnamese institution.
Workshop papers (15)
Workshop papers found on OpenReview and papers placed in Findings, for reference; not counted, except that a Findings paper counts as a main-track rejection (see About).
-
ICLR 2026 · workshop Trustworthy AI,
BackFed: A Standardized and Efficient Benchmark Framework for Backdoor Attacks in Federated Learning
-
ICML 2026 · workshop AdaptFM
SparseSAM: Structured Sparsification of Activations in Segment Anything Models
-
ICML 2026 · workshop AdaptFM
StructSAM: Structure- and Spectrum-Preserving Token Merging for Segment Anything Models
-
ICML 2026 · workshop FM4LS
Structural Bottleneck Reasoning: Efficient Medical VQA via Concept Alignment
-
ICML 2026 · workshop AI4GOOD,
Understanding Consistency Through Internal Representations in Large Vision-Language Models
-
CVPR 2026 · Findings
Onboarding Without Forgetting: Hypernetwork Personalization with Data-Free Replay for Personalized Federated Learning
-
ICLR 2025 · workshop WMARK,
SpARK: An Embarrassingly Simple Sparse Watermarking in LLMs with Enhanced Text Quality
-
ICLR 2025 · workshop Data Problems
Synthesizing Physical Backdoor Datasets: An Automated Framework Leveraging Deep Generative Models
-
NeurIPS 2025 · workshop Reliable ML,
Clean-Label Physical Backdoor Attacks with Data Distillation
-
ICML 2025 · workshop GenBio
From Fragments to Geometry: A Unified Graph Transformer for Molecular Representation from Conformer Ensembles
-
ICML 2025 · workshop ES-FoMo-III,
How Many Tokens Do 3D Point Cloud Transformer Architectures Really Need?
-
ICML 2025 · workshop MUGen
Unveiling Concept Attribution in Diffusion Models
-
ICML 2025 · workshop DIG-BUG,
Weak-to-strong Generalization via Formative Learning from Student Demonstrations & Teacher Evaluation
-
ACL 2024 · Findingsunofficial
Fooling the Textual Fooler via Randomizing Latent Representations
Unofficial: listed on their own page; not in the official list yet.
-
NeurIPS 2023 · workshop BUGS
Clean-label Backdoor Attacks by Selectively Poisoning with Limited Information from Target Class