Analytical Study of Momentum-Based Acceleration Methods in Paradigmatic High-Dimensional Non-Convex Problems.
Stefano Sarao MannelliPierfrancesco UrbaniPublished in: NeurIPS (2021)
Keyphrases
- high dimensional
- analytical methods
- empirical studies
- practical problems
- high dimensionality
- methods can be applied
- significant improvement
- qualitative and quantitative
- high dimensional problems
- data sets
- alternative methods
- search methods
- machine learning methods
- np complete
- low dimensional
- computational cost
- convex optimization problems
- estimation problems
- high dimensional spaces
- high dimensional data
- preprocessing
- feature selection
- neural network