705.801.21 - Independent Study in Artificial Intelligence I

Expanded Course Description

This independent study, conducted under EN.705.801 Independent Study in Artificial Intelligence, focuses on a research problem of current relevance to artificial intelligence and public policy analytics. The purpose of the study is to enable graduate-level inquiry into label-efficient text representation learning, with specific application to public policies across the AI--energy industrial chain. Consistent with the expectations of EN.705.801, the work will culminate in a substantial research paper and accompanying technical project demonstrating mastery of advanced AI methodologies.

The proposed research seeks to examine how high-quality policy document embeddings may be constructed under constrained expert labeling resources. The inquiry is motivated by a growing need for scalable analytic tools that can support policy analysis, strategic planning, and decision-making in domains that integrate artificial intelligence, energy systems, and industrial development. The study aims to contribute methodologically and empirically to the literature on weak supervision, representation learning, and AI-assisted policy analysis.

Course Structure

• Weeks 1--2: Finalization of metadata schema; implementation of ingestion, chunking, and citation anchoring; creation of an initial retrieval evaluation set; selection and validation of the expert-labeled gold subset.
• Weeks 3--5: Development of LLM-driven weak labels and heuristics for retrieval facets (doc nature/authority/topic/section roles) and policy attributes; implementation of faceted retrieval / query routing; completion of initial ablations.
• Weeks 6--10: Comparison of baseline encoders vs lightly fine-tuned encoders; evaluation of semantic search and retrieval metrics; completion of clustering/attribute prediction primarily as auxiliary analyses and error diagnostics; quantifation of marginal value of expert labels.
• Weeks 11--14: Paper writing, refinement, and final project preparation.

Course Topics

The project will evaluate two primary modeling pathways:

• Baseline Encoders: evaluation of state-of-the-art pretrained sentence and document embedding models without domain-specific fine-tuning, such as Sentence-BERT and contemporary transformer embedding models.
• Label-Efficient Encoders: development of lightly fine-tuned models trained on weak signals that primarily support retrieval—document facets, inferred section roles, and policy attributes (e.g., "green" conditions, "clawback" provisions) with optional refinement using expert annotations

The effectiveness of these representations will be assessed through three downstream tasks that reflect realistic analytic workflows:

• Semantic Search: retrieval of policy segments in response to domain-relevant analyst queries.
• Clustering and Typology Formation: identification of emergent policy categories and characterization of structural similarities across jurisdictions.
• Attribute Prediction: classification of policy provisions derived from embedding features.

Weak supervision will rely on structured prompting of large language models, combined with rule-based heuristics for identifiable patterns (e.g., incentive structures, reporting requirements). Labels will be applied selectively where they support retrieval (facets and query routing) and evaluation (stratified analysis of errors). Experiments will quantify the marginal value of expert labels by comparing retrieval and downstream task performance under (i) no labels, (ii) weak labels only, and (iii) weak + expert refinement.

The study will evaluate a hybrid evidence-retrieval design combining a curated, versioned snapshot corpus used for reproducible experiments, optionally augmented with targeted web search for recency and long-tail sources (e.g., Tavily-style web search APIs designed for RAG/agentic workflows). Web-retrieved sources may be cached with provenance metadata and assigned an authority tier to preserve traceability and avoiding curating the same documents multiple times.

Early in the semester, the project will inventory candidate external datasets and repositories relevant to AI–energy infrastructure analysis (e.g., commercial real estate databases, grid interconnection queue datasets, power market pricing/forecast data, and data-center market intelligence). For each candidate source, the proposal will document access model, estimated cost, and feasibility within the semester. These resources are treated as optional extensions.

Course Goals

At the conclusion of the independent study, the student will be able to:

• Develop and curate a versioned policy text corpus and metadata schema designed to support retrieval, citation, and reproducible evaluation.
• Apply and compare state-of-the-art transformer-based text embedding models using both off-the-shelf and lightly fine-tuned approaches.
• Develop and evaluate weak supervision pipelines that combine large language model (LLM) outputs with heuristic labeling rules.
• Conduct formal evaluations of embedding models using established metrics for semantic search, clustering coherence, and attribute classification.
• Analyze trade-offs between labeling effort and retrieval/model performance, identifying where expert annotations provide the greatest benefit.
• Produce a conference- and journal-quality research manuscript as the primary scholarly outcome of the independent study, integrating methodology, experiments, results, and implications, with a staged dissemination plan that supports iterative refinement and academic rigor:
  
  • Stage 1: Professional Validation: Present preliminary research results at the Military Operations Research Society (MORS) 94th Symposium to obtain practitioner-oriented feedback and assess operational relevance.
  • Stage 2: Peer-Reviewed Conference Submission (August 2026): Expand and refine the manuscript based on symposium feedback and submit to one of the following venues:
    
    • SPIE Defense + Commercial Sensing 2027, emphasizing applied defense analytics, sensing, and mission-driven AI methodologies, or
    • NeurIPS, emphasizing novel AI/ML contributions, agentic architectures, learning dynamics, or theoretical foundations.
  • Stage 3: Archival Journal Publication: Produce a substantially extended and rigorously evaluated version of the work and submit it to a leading peer-reviewed AI journal (e.g., the Journal of Artificial Intelligence Research), incorporating additional experiments, ablation studies, and formalized analysis.

Textbooks

Lecture notes and literature review

Other Materials & Online Resources

Course Notes

Required Software

Python
Hugging Face
GitHub
Other

Student Coursework Requirements

To satisfy the requirements of EN.705.801, the student will submit:

• A substantial research paper suitable for submission to a peer-reviewed venue such as IEEE or SPIE, presenting the research problem, methodology, experiments, results, and conclusions.
• A Python-based technical project, including code for corpus construction, labeling, embedding experiments, and evaluation tasks.
• Supporting materials, including:
  
  • Figures, tables, and data-processing notes required for reproducibility.
  • A comprehensive extbf{Model Description Document (MDD)} or extbf{Algorithm Description Document (ADD)} that formally specifies model architectures, training procedures, assumptions, data flows, and evaluation protocols. This document should follow recognized standards for technical model disclosure and support the analytical rigor of the study.

These deliverables collectively fulfill the course requirement of producing a significant AI-related paper or project.

Grading Policy

Score Range	Letter Grade
100-97	= A+
<97-93	= A
<93-90	= A−
<90-87	= B+
<87-83	= B
<83-80	= B−
<80-77	= C+
<77-73	= C
<73-70	= C−
<70-67	= D+
<67-63	= D
<63	= F

Academic Policies

Deadlines for Adding, Dropping and Withdrawing from Courses

Students may add a course up to one week after the start of the term for that particular course. Students may drop courses according to the drop deadlines outlined in the EP academic calendar (https://ep.jhu.edu/student-services/academic-calendar/). Between the 6th week of the class and prior to the final withdrawal deadline, a student may withdraw from a course with a W on their academic record. A record of the course will remain on the academic record with a W appearing in the grade column to indicate that the student registered and withdrew from the course.

Academic Misconduct Policy

All students are required to read, know, and comply with the Johns Hopkins University Krieger School of Arts and Sciences (KSAS) / Whiting School of Engineering (WSE) Procedures for Handling Allegations of Misconduct by Full-Time and Part-Time Graduate Students. This policy prohibits academic misconduct, including but not limited to the following: cheating or facilitating cheating; plagiarism; reuse of assignments; unauthorized collaboration; alteration of graded assignments; and unfair competition. Course materials (old assignments, texts, or examinations, etc.) should not be shared unless authorized by the course instructor. Any questions related to this policy should be directed to EP’s academic integrity officer at ep-academic-integrity@jhu.edu.

Students with Disabilities - Accommodations and Accessibility

Johns Hopkins University is committed to providing welcoming, equitable, and accessible educational experiences for all students. If disability accommodations are needed for this course, students should request accommodations through Student Disability Services (SDS) as early as possible to provide time for effective communication and arrangements.  For further information about this process, please refer to the SDS Website.

Student Conduct Code

The fundamental purpose of the JHU regulation of student conduct is to promote and to protect the health, safety, welfare, property, and rights of all members of the University community as well as to promote the orderly operation of the University and to safeguard its property and facilities. As members of the University community, students accept certain responsibilities which support the educational mission and create an environment in which all students are afforded the same opportunity to succeed academically.  For a full description of the code please visit the following website: https://studentaffairs.jhu.edu/policies-guidelines/student-code/

Classroom Climate

JHU is committed to creating a classroom environment that values the diversity of experiences and perspectives that all students bring. Everyone has the right to be treated with dignity and respect. Fostering an inclusive climate is important. Research and experience show that students who interact with peers who are different from themselves learn new things and experience tangible educational outcomes. At no time in this learning process should someone be singled out or treated unequally on the basis of any seen or unseen part of their identity.    If you have concerns in this course about harassment, discrimination, or any unequal treatment, or if you seek accommodations or resources, please reach out to the course instructor directly. Reporting will never impact your course grade. You may also share concerns with your program chair, the Assistant Dean for Diversity and Inclusion, or the Office of Institutional Equity. In handling reports, people will protect your privacy as much as possible, but faculty and staff are required to officially report information for some cases (e.g. sexual harassment).

Course Auditing

When a student enrolls in an EP course with “audit” status, the student must reach an understanding with the instructor as to what is required to earn the “audit.” If the student does not meet those expectations, the instructor must notify the EP Registration Team [EP-Registration@exchange.johnshopkins.edu] in order for the student to be retroactively dropped or withdrawn from the course (depending on when the "audit" was requested and in accordance with EP registration deadlines). All lecture content will remain accessible to auditing students, but access to all other course material is left to the discretion of the instructor.