Learning path
Nodes3
Raw rows shared0
Rounds5

Round history

RoundAccuracyLoss
10.6860.630
20.6860.620
30.6860.617
40.6860.616
50.6860.615

Final weights

bias +0.416
pathogen_CRAB +0.296
pathogen_CRE +0.114
pathogen_CRPA +0.006
antibiotic_meropenem +0.101
antibiotic_colistin +0.087
antibiotic_amikacin +0.111
antibiotic_cefiderocol +0.118
icu +0.250
genomics_available +0.053

Module README

03 Basic Federated Learning

Minimal federated training with standard Python.

Didactic objective: predict whether a synthetic row will be resistant using simple features:

  • Pathogen.
  • Antibiotic.
  • ICU yes/no.
  • Genomics available yes/no.

Each node trains a local logistic regression for a few iterations. The coordinator receives weights and example counts, and computes a weighted average.

This is not a validated scientific model. It is designed to understand the federated learning pattern.

Run

From Desarrollo:

python .\03_federated_learning_basico\federated_learning_minimo.py