Optimized the subspace cluster detection function of the DRESS algorithm to enhance both execution time and the quality of subspace clusters.
- Analyzed the existing implementation of the DRESS algorithm to identify performance bottlenecks.
- Implemented parallel processing techniques to leverage multi-core processors and reduce execution time.
- Enhanced the cluster detection logic by proposing a weighted distance similarity score instead of using only a distance similarity score, improving the accuracy and relevance of identified subspace clusters.
- Proposed method improved the quality of subspace clusters ensuring more meaningful and actionable results.