Finding autoencoders useful for performance anomaly root cause analysis. Observation: while other methods help identify the anomaly (what, when), the autoencoder algorithm has strong utility for triaging and finding the root-cause needle in the thousands-of-metrics haystack.
South Australia's electricity generation, yesterday 1.30pm:
Solar PV > 100% of total demand
Solar+Wind > 130%
Gas 6% for stability
Excess exported
3 days at >100% renewables in a 1.5GW system.
It's a glimpse of the future.
https://t.co/kxEdGZtCaZ
Buenos días #EnfermosdeF1,
Este fin de semana nos vamos de #f1esta al Autódromo Hnos. Rodríguez. GP de México 2021
Neumáticos elegidos: C2, C3 y C4 (vía @pirellisport)
Sin cambios respecto 2019. El año pasado no hubo carrera por la Covid.
#F1#MexicoGP#TechF1#F12021
Hand assembling my prototype 100A BMS for initial testing. Next up: training a TinyML model on cell statistics to accurately predict capacity under varied load conditions. #TinyML#Esk8