New #TMLR-Paper-with-Video:

Socrates Loss: Unifying Confidence Calibration and Classification by Leveraging the Unknown

Sandra Gómez-Gálvez, Tobias Olenyi, Gillian Dobbie, Katerina Taskova

https://tmlr.infinite-conf.org/paper_pages/DONqw1KhHq

#overfitting #accuracy #neural

Here is where i come from:
If you train your machine on data with too few dimensions, you cannot capture the behaviour.
When learning statistics, we are taught of the perils of #overfitting: If as many dimensions as data points, predictions tend to perform worse on out-of-sample data.

Here are considerations for statisticians and scientists, about system thinking and onboarding context 🧵

#liveability #Cybernetics #inference #modeling #modelling #complexity #systems #generalization #statistics #models #ML #evolvability #doubleDescent #overfitting #overparameterization #stats […]

Here are considerations for statisticians and scientists, about system thinking and onboarding context 🧵

#liveability #inference #modeling #modelling #complexity #systems #generalization #statistics #models #ML #evolvability #doubleDescent #overfitting #stats #probabilities #feedback

⚖️ El Equilibrio del Machine Learning
Sesgo alto = underfitting. Varianza alta = overfitting. El punto óptimo combina bajo sesgo y baja varianza. Y entre dos modelos iguales, el más simple gana.
#MachineLearning #SesgoVarianza #Overfitting #Underfitting #NavajaDeOckham #CienciaDeDatos