https://reallyacepeter.github.io/ Explorations in Artificial Intelligence, System Architecture, and Computational Science. By Ace Lee, Researcher & Tech Entrepreneur. 2026-06-01T20:00:26+09:00 Dong Jun Lee https://reallyacepeter.github.io/ Jekyll © 2026 Dong Jun Lee /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png 2026-05-26T09:00:00+09:00 2026-05-26T09:00:00+09:00 https://reallyacepeter.github.io/posts/Why-Similarity-Search-Requires-Representation-Learning/ Dong Jun Lee

Introduction In the previous post, I discussed why forensic dental AI depends on more than just models. Before a system can learn anything useful, it must first have access to: Structured dental records Reliable data governance Expert-reviewed information Trustworthy data pipelines However, another challenge remains. Even with a well-designed database, how can a system identify r...

2026-05-19T09:00:00+09:00 2026-05-19T09:00:00+09:00 https://reallyacepeter.github.io/posts/The-Real-Bottleneck-in-Forensic-Dental-AI-Data,-Governance,-and-Trust/ Dong Jun Lee

Introduction In the previous post, I discussed why forensic dental identification cannot rely only on exact queries. A missing person identification system does not operate in a clean database environment. In real forensic scenarios: Ante-mortem records may be incomplete Post-mortem findings may be degraded Dental conditions may change over time Imaging data may be noisy or distort...

2026-04-20T14:00:00+09:00 2026-04-20T14:00:00+09:00 https://reallyacepeter.github.io/posts/From-Query-to-Match-Designing-a-Similarity-Engine-for-Dental-Identification/ Dong Jun Lee

Introduction Most data systems are built around queries. Input a condition Retrieve matching records This paradigm works well when the data is complete and the queries are precise. However, forensic identification does not operate under such conditions. In real-world scenarios: Data is incomplete Observations are noisy Records are often inconsistent across time This leads t...

2026-04-11T21:10:00+09:00 2026-04-11T21:10:00+09:00 https://reallyacepeter.github.io/posts/Evaluating-a-Forensic-Dental-Identification-System-From-Data-Integrity-to-System-Validation/ Dong Jun Lee

Introduction Building a system is only half the problem. In forensic identification, correctness is not optional. A system that stores dental records is not useful unless it can be trusted under real-world conditions. This raises a critical question: How do we validate that a forensic dental identification system actually works? This post explores the evaluation and validation of a mob...

2026-04-07T09:00:00+09:00 2026-04-07T20:15:33+09:00 https://reallyacepeter.github.io/posts/The-Cost-of-Over-Optimization-A-Decision-Making-Lesson-from-a-Real-World-Tradeoff/ Dong Jun Lee

Introduction In many real-world systems, decision-making is framed as an optimization problem. The objective appears straightforward: Minimize loss Maximize outcomes However, in practice, this approach often produces unintended results. Optimizing individual decisions does not always lead to optimal system behavior. This phenomenon—referred to here as over-optimization—occurs wh...