Recod from Social Media – #247

From Anderson Rocha on Linkedin:

How can we solve a #classification problem #WithoutLabels ? Check out the latest result of our Artificial intelligence Lab., Recod.ai ( http://Recod.ai ).

General self-supervised learning methods fail to learn discriminative features when classes have closer semantics. They are also often devised for specific modalities.

We propose a strategy to tackle this problem and enable learning from unlabeled data (from different modalities) even when samples from different classes are not prominently diverse. We leverage a novel #EnsembleBasedClustering strategy where clusters derived from different configurations are combined to generate a better grouping for the data samples in a  #FullyUnsupervised  way and without hyper-parameter tuning.

Check out our work “Reasoning for Complex Data through Ensemble-based Self-Supervised Learning” on ArXiv:  https://lnkd.in/emZf34tK  🙂

This is joint work with  Gabriel Bertocco , Antonio Theophilo and Fernanda Andalo, Ph.D. !

#ai   #ml   #deeplearning   #selfsupervised   #reid #authorshipattribution   #recodai   #unicamp #artificialintelligence #data