Author Archive

Expert Data Scientist at @synergicpartner | PhD in Robotics/Machine Learning. Follow me on Github (smorante)

Warnings about normalizing data
  • Santiago Morante
  • Algorithms . Algoritmos . Analítica Avanzada . Big Data . Blog Synergic . Data . Data Normalization . Data Science . datos . Machine Learning . Normalización de Datos .

For many machine learning algorithms, normalizing the data of analysis is a must. A supervised example would be neural networks. It is known that normalizing the input data to the networks improve the results. If you don't believe me it's ok (no offense taken), but you may prefer to believe Yann Le Cunn (Director of AI Research in Facebook and founding father of convolutional networks) by checking section 4.3 of this paper. You can catch up the idea with the first sentence of the section: Convergence [of backprop] is usually faster if the average of each input variable over the training set is close to zero. Among other things, one reason is that when the neural network tries to correct…

Featured customer