Ionel Eduard STAN, Ph.D. / University of Milano-Bicocca

Ionel Eduard STAN, Ph.D.

Intelligent Sensing Laboratory (ISLab)Department of Informatics, Systems and Communication (DISCo)University of Milano-Bicocca44 publications · 2 editorial boards
Ionel Eduard STAN, Ph.D., photographed against a plain background
I am an Assistant Professor (Ricercatore a Tempo Determinato di tipo A—RTD/A) at the University of Milano-Bicocca, where I bridge the gap between formal methods, symbolic reasoning, and cutting-edge AI technologies.
Ionel Eduard STAN, Ph.D. · Intelligent Sensing Laboratory (ISLab)

Quoted verbatim from content/cv.yaml · profile.bio.long, first sentence.

The workcontent/cv.yaml · profile.bio.long

I am an Assistant Professor (Ricercatore a Tempo Determinato di tipo A—RTD/A) at the University of Milano-Bicocca, where I bridge the gap between formal methods, symbolic reasoning, and cutting-edge AI technologies. My expertise lies in leveraging formal logic and symbolic AI to enhance the interpretability and transparency of machine learning models, particularly within the rapidly evolving field of formal explainable AI (FXAI).

At the Intelligent Sensing Lab (ISLab), I collaborate with a team of deep learning experts, bringing my background in symbolic methods to bear on the challenges of explainability. Our goal is to ensure that the AI systems we develop are effective and trustworthy, particularly in high-stakes applications like healthcare. This work is central to our contributions to the ANTHEM project, where we focus on developing smart environments and innovative sensors for proximity medicine, working alongside leading institutions and healthcare providers.

In addition to my research, I am deeply committed to my role as an educator. I actively seek out and mentor students, guiding them through their theses and fostering their development as future researchers. My teaching is informed by my belief in the importance of transparency and ethics in AI, principles that are increasingly critical as AI technologies become more integrated into society.

A vision of AI drives my work as a tool for enhancing human capabilities, built on a foundation of trust, ethics, and collaboration. I am excited to continue exploring the potential of neuro-symbolic AI and FXAI, and I am always open to new partnerships that align with these goals. Let’s connect and push the boundaries of what AI can achieve together.

content/cv.yaml · profile.bio.long — 4 paragraphs, rendered verbatim. The printed CV does not carry them: it prints profile.bio.short, the same person in the third person.

Three strandsweb/src/lib/strands.ts

  1. Interpretable AI

    Formal logic and symbolic AI used to make learned models inspectable: temporal decision trees and forests, rule extraction, and explanations that can be checked rather than taken on trust.

  2. Temporal and logic methods

    Modal and interval temporal logic as a learning language — tableau systems, fuzzy generalisations, model checking, and modal symbolic learning over time series, audio and sensor streams.

  3. Agentic AI

    Systems that plan and act, not only predict. Taught rather than published so far: it runs through two Bicocca courses, and the nearest work in the record is on strategic reasoning and automated synthesis for multi-agent systems.

Unlike everything else on this page, the strands are not read out of a repository file — nothing in the repository states a strand structure. The copy lives in web/src/lib/strands.ts, and each strand names the files its claim rests on.

Interpretable AI — content/cv.yaml, content/publications.bib

Temporal and logic methods — content/publications.bib

Agentic AI — content/cv.yaml

The countcontent/publications.bib

  • 44entries in this bibliography
  • 21conference entries
  • 8journal entries
  • 7workshop entries
  • 6under review entries
  • 2software entries
  • 2editorial boards
  • 15service roles on record

content/publications.bib — 44 entries, 2017–2026, counted at build by parsing the file. The two service figures are counted from content/cv.yaml: 2 entries of service[] whose title matches “editor” (Elsevier Neurocomputing Journal, Frontiers in Artificial Intelligence Journal), and 15 entries in that list altogether.

Associate Editorcontent/cv.yaml · service[] · role: Associate Editor

Technical Committee Membercontent/cv.yaml · service[] · role: Technical Committee Member

Tutorials/Special Session Co-Chaircontent/cv.yaml · service[] · role: Tutorials/Special Session Co-Chair

Area Chaircontent/cv.yaml · service[] · role: Area Chair

Program Committeecontent/cv.yaml · service[] · role: Program Committee

Reviewercontent/cv.yaml · service[] · role: Reviewer

content/cv.yaml — service[], 15 entries grouped into 6 roles by their own title field (Associate Editor: 2, Technical Committee Member: 1, Tutorials/Special Session Co-Chair: 1, Area Chair: 2, Program Committee: 4, Reviewer: 5), in the order those roles first appear in the file. The meta on each row is the entry's dates where it states a term and its years[] where it states editions; the badges are the metric field, linked to the rank_url beside it.

This column and /professional_activities/ read the same list through the same grouping in web/src/lib/cv-schema.ts, so they cannot disagree. They once came from two files that had drifted apart; there is one file now, and it is the one the printed CV is generated from.

web/src/lib/consistency.ts — the consistency gate: 1 check · 18 comparisons · 0 contradictions · 1 excused · 2 records with nothing to compare against. Each check compares two hand-typed records of one fact — a date the fact states and the date it says it was announced on — joined by being the same entry rather than by matching prose. A contradiction refuses astro build; astro dev never fails, so this block is where the author sees one.

announced-in-own-year — an announcement date must fall in the year of the fact it hangs on, or the year before it (invited in N−1 to serve at the N edition). Over content/cv.yaml, content/publications.bib.

announced-in-own-year — excused · dates Mar 2024–Present vs announced 2025-03-03T10:00:00+02:00 · service[1] "Frontiers in Artificial Intelligence Journal" · content/cv.yaml:240, content/cv.yaml:239
Appointed to the board in March 2024 and announced on the site a year later, on 2025-03-03, when the special issue launched. Both dates are correct and both are stated here. · expires 2027-01-01

Not compared — service[10] "Elsevier Engineering Applications of Artificial Intelligence Journal" carries an announcement date but states no year of its own to compare it against (content/cv.yaml:324).

Not compared — service[11] "IEEE Journal of Biomedical and Health Informatics" carries an announcement date but states no year of its own to compare it against (content/cv.yaml:330).

Latelycontent/cv.yaml

All 70 items →

70 announcements, generated from the facts themselves — content/cv.yaml, content/publications.bib, content/talks.bib, content/posts/**/*.{md,mdx}. Nothing is written twice: a fact is announced on the date it carries, and the feed shows that date at the precision the source states it and no finer.

This column is capped at 6, so 6 of 70 are shown here and 64 are not; the full feed is at /lately/.