Course policies
In brief
Submit work you can explain and verify. Do not cheat, plagiarize or fabricate results. Credit reused text, code, data and models. Use collaboration and AI tools within the rules of the activity. Respect data and evaluation restrictions. Follow the rules on each competition page and examination paper.
Academic integrity
Present your own contribution honestly. Copying or lightly rewriting another person’s text, code, figures or analysis without attribution is plagiarism. Attribution is still needed for adapted material. A citation does not make reuse permissible when the activity requires independent work.
Do not impersonate another student, submit someone else’s work as your own, share assessed answers where collaboration is prohibited, or obtain unauthorized assistance during an exam.
Reuse, collaboration and AI
When reuse is permitted, identify the original source and make your contribution clear. Credit datasets, pretrained models, libraries and substantial borrowed code; respect their licences. Describe substantive assistance, including AI tools, sufficiently to distinguish the assistance from your own decisions. Check every generated reference before citing it.
The activity instructions determine permitted collaboration, team size, tools, external data and pretrained models. AI coding agents explains the course’s starting point; competition-based labs explains where to find competition-specific conditions. A tool’s availability does not establish permission to use it in an assessment.
Honest and reproducible experiments
Report the experiments you actually ran. Do not invent measurements, hide test-set use, alter evidence or claim an unperformed experiment. Explain data splits, preprocessing, model selection and evaluation, including important limitations or unsuccessful results that affect your conclusion.
Keep enough code, settings and instructions to let another person check your main results, within the activity’s data-sharing restrictions. Be ready to explain and reproduce your submitted work. A leaderboard score alone does not demonstrate understanding.
Data and evaluation
Respect licences, privacy and any restrictions on uploading or sharing data. Follow the competition’s rules for test data, external sources, submissions and teams. Do not use leaked labels or private evaluation information to bypass the intended task.
Competitions and exams
Each competition page states its specific conditions, including permitted data, architectures, tools, teamwork and submissions. Each examination paper states the conditions for that exam, including permitted aids and collaboration. Read those instructions before starting. If a requirement is unclear or appears inconsistent, ask the instructor before relying on an interpretation.
The syllabus gives the assessment structure and points. This policy does not introduce additional penalties, deadlines or grading weights.
Further guidance
The emphasis on verifiable claims, transparent experiments and attribution is informed by the NeurIPS Paper Checklist, code submission guidance and academic-integrity guidance. These are research references; the course’s activity instructions govern student assessments.
TODO(instructor): Add or link the institution’s authoritative accessibility, academic-integrity, privacy, late-work and student-support policies, including the applicable procedure and consequences for suspected misconduct. No institutional process is invented here.