LSALSA: accelerated source separation via learned sparse coding

Computer Science

• Component Analysis 100%
• Source Separation 100%
• Gradient Descent 66%
• Alternating Direction Method of Multipliers 66%
• Theoretical Framework 33%
• Optimization Problem 33%
• Efficient Algorithm 33%
• Sparse Representation 33%
• Analysis Framework 33%
• Steepest Descent 33%
• Reconstruction Error 33%
• Gradient Direction 33%

Keyphrases

• Salsa 100%
• Morphological Component Analysis 50%
• Split Augmented Lagrangian Shrinkage Algorithm 33%
• Loss Landscape 33%
• Multiple Dictionaries 16%
• Alternating Optimization 16%
• Sparse Combination 16%
• Quadratic Reconstruction 16%
• Dictionary Matrix 16%