Transformes for Time Series: Is the New State of the Art (SOA) Approaching? - Ezequiel Lanza, Intel
Ezequiel will embark us on a tour of the well-known transformers that are beginning to show their fruits in an environment such as TIME SERIES. What are the challenges and why is it so difficult to use the same architectures that are used for NLP? A tour of the most significant opensource implementations and papers (Informer, spacetimeformer) to finally arrive on an implementation on latency prediction in a Kubernetes cluster vs current SOA showing their pros and cons and explaining where we are now in the The presentation will not go deeper in the technical aspects.