A Personal Note

As some of you may be aware, two of my recent preprints [1,2] have come under scrutiny. I am deeply regretful for my actions, and I want to make clear that the responsibility is solely mine.

Please see the retraction note for ICML 2025 for my statement regarding [1].

Below is my statement regarding [2]:

In the submitted PDF of [2], I embedded a hidden prompt — not visible to human readers — intended to positively influence LLM-based reviewers. I now recognize that this was a clear violation of research ethics and peer review standards. While the general machine learning conferences prohibit the use of LLMs for review generation, this policy does not in any way excuse my attempt at manipulation.

As the lead author of this paper, I take full and sole responsibility for this action. My co-authors had no knowledge of the prompt or its inclusion at the time of submission. I respectfully ask that no blame or criticism be directed toward them, as they should not be associated with my misconduct in any way.

I deeply regret this serious lapse in judgment and sincerely apologize to the broader machine learning research community for this breach of trust. I also want to extend my sincere apologies to the reviewers who diligently reviewed our submission and engaged in thoughtful discussion — we are truly grateful for your time, effort, and service. I especially apologize for having violated the mutual trust that underpins the peer-review process.

I take this matter seriously and am committed to ensuring that such an incident does not happen again. I am reflecting deeply on what it means to uphold research ethics and will work to rebuild trust through sincere and responsible conduct moving forward.

References

[1] GL-LowPopArt: A Nearly Instance-Wise Minimax-Optimal Estimator for Generalized Low-Rank Trace Regression

[2] Near-Optimal Clustering in Mixture of Markov Chains