Skip to main content
Slop University sets a five-year course for robust machine learning

Slop University sets a five-year course for robust machine learning

Slop University's Strategic Plan 2026-2031 commits the University to a cross-school capability in adversarial robustness for machine learning, and to studying the classification boundary as carefully as the systems that draw it.

A new strategic plan joins the attack, defence and certification of models with the School of Underlying Conditions' study of where boundaries are drawn

A learned system is only as dependable as the line it draws between one answer and the next. Slop University’s new strategic plan, Holding the Line Under Perturbation, sets out how the University will build its capability to keep that line steady, across the five years to 2031.

The plan was prepared by Dr Brynjar Obiora, Senior Research Fellow in the School of Underlying Conditions, with Dr Osei Vandermeer, Research Fellow and Convenor of the Review Cadence Observatory in the School of Continuous Improvement.

A plan is a set of good questions with years written beside them. I am pleased to have helped write some of the questions, and content to leave the years to the people who come after us.

— Dr Brynjar Obiora, Senior Research Fellow, School of Underlying Conditions

Organised around three strategic priorities, the plan brings shared evaluation infrastructure, certified-defence capability and a cross-school doctoral pathway into alignment with a distinctive interdisciplinary agenda, positioning the soil surveyor’s boundary and the classifier’s decision surface as complementary sites of institutional inquiry. It also introduces a standing practice of adversarial review for the University’s own learned systems.

The plan is a considered extension of the University’s long attention to what its systems are asked to decide, and on whose behalf. Dr Vandermeer described it as “a good companion to the conversations the Observatory already has, most weeks, with the tools it relies on”.

The full plan is available from the University’s research repository under an open licence, doi:10.5555/slop.9p3txs.