Better Mini-Batch Algorithms via Accelerated Gradient Methods
Andrew CotterOhad ShamirNathan SrebroKarthik SridharanPublished in: CoRR (2011)
Keyphrases
- significant improvement
- computational cost
- benchmark datasets
- methods outperform
- computer vision algorithms
- computationally expensive
- synthetic and real datasets
- problems in computer vision
- empirical studies
- optimization methods
- learning algorithm
- steepest ascent
- machine learning methods
- machine learning algorithms
- optimization problems
- search methods
- computationally complex
- theoretical analysis
- social networks
- computationally intensive
- methods require
- alternative methods
- computationally efficient
- data structure
- heuristic methods
- genetic algorithm
- batch mode
- online algorithms
- high computational complexity
- image processing algorithms
- hybrid method
- exhaustive search
- mathematical models
- evaluation metrics
- parameter settings
- high dimensional data