Dimensions overall score 3.0

GitHub Code Pulse

No public code linked for this paper yet.

• Deep ensemble learning with transfer learning significantly improves classifier performance for bipolar disorder and schizophrenia classification compared to simple machine learning.result 90%
• Deep ensemble learning reaches a performance plateau when 10 models are included in the ensemble for BD and SCZ classification.result 85%
• Transfer learning induces better generalization in classification models for psychiatric disorders.technical 80%
• Using pre-trained models constrains transfer learning models to stay in the same basin of the loss function.technical 85%
• Deep learning models with randomly initialized weights do not stay in the same basin of the loss function.technical 85%
• Transfer learning and deep ensemble learning reduce the variability of single-subject classification models in BD and SCZ classification.result 80%
• The study investigated training stability by comparing multiple trainings with the same backbone but different initializations.method 90%
• The study accounts for epistemic uncertainty associated with uncertainty in model parameter estimation.method 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card