Mauricio Roman MauricioRoman

## non_degenerate_mlp_gram.py
"""Empirical evaluation of the extended feature Gram matrix of a ReLU MLP

Here we try to estimate the spectrum of the H^\infty matrix as defined in:

Gradient Descent Provably Optimizes Over-parameterized Neural Networks (2018)
Simon S. Du, Xiyu Zhai, Barnabas Poczos, Aarti Singh
https://arxiv.org/abs/1810.02054

Theorem 4.1 relies on the assumption that H^\infty has a strictly positive
minimum eigenvalue. The following computes an estimate of this eigenvalue

## index.md

      
        
          
            
              
              1 file
            
          
          
            
              
              4 forks
            
          
          
            
              
              23 comments
            
          
          
            
              
              44 stars
            
          
        
        
          
              
          
          
            
                johan
                / index.md
            
            
              Last active
              September 26, 2023 16:35
            
              
                osx + java 7 = painfully easy
              
          
        
      
        
  
      
    Step 1

Does your osx terminal speak java 7? Start Terminal.app and try: java -version:
> java -version
java version "1.7.0_51"
Java(TM) SE Runtime Environment (build 1.7.0_51-b13)
Java HotSpot(TM) 64-Bit Server VM (build 24.51-b03, mixed mode)
	"""Empirical evaluation of the extended feature Gram matrix of a ReLU MLP

	Here we try to estimate the spectrum of the H^\infty matrix as defined in:

	Gradient Descent Provably Optimizes Over-parameterized Neural Networks (2018)
	Simon S. Du, Xiyu Zhai, Barnabas Poczos, Aarti Singh
	https://arxiv.org/abs/1810.02054

	Theorem 4.1 relies on the assumption that H^\infty has a strictly positive
	minimum eigenvalue. The following computes an estimate of this eigenvalue