Worse than Random: The Importance of a Baseline for Unsupervised Feature Selection
A recent paper highlights the necessity of establishing a baseline for evaluating unsupervised feature selection methods. The authors propose using random feature selection as a benchmark, revealing that many advanced methods do not outperform this baseline. This emphasizes the need for consistent improvement over random selection in future developments.
- ▪The paper suggests that many unsupervised feature selection methods are evaluated without a proper baseline.
- ▪Using random feature selection as a baseline can help assess the effectiveness of new methods.
- ▪The authors found that some state-of-the-art methods were outperformed by random selection in terms of performance and efficiency.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Machine Learning arXiv:2605.22973 (cs) [Submitted on 21 May 2026] Title:Worse than Random: The Importance of a Baseline for Unsupervised Feature Selection Authors:Muhammad Rajabinasab, Michael E. Houle, Oussama Chelly, Arthur Zimek View a PDF of the paper titled Worse than Random: The Importance of a Baseline for Unsupervised Feature Selection, by Muhammad Rajabinasab and 3 other authors View PDF HTML (experimental) Abstract:Many novel unsupervised feature selection methods are proposed each year, yet their empirical evaluation is limited to supervised and unsupervised evaluation metrics computed on selected datasets, along with comparisons to existing methods.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
