In her talk at ODSC West 2024, Professor Yan Liu from USC explored how foundation models, inspired by GPT architectures, can fundamentally change the analysis of time series data. With an ever-increasing volume of real-time data from various sectors like climate science, healthcare, and finance, Liu emphasized the urgent need to use AI not just for efficiency but for addressing crucial societal issues. She detailed the unique characteristics of time series data, including its dynamics, noise, and physical grounding, necessitating innovative modeling approaches that respect these nuances and reflect reality more closely than other data types.
Foundation models, inspired by GPT architectures, have the potential to revolutionize our understanding and application of complex time series data in various fields.
These models offer a unique opportunity to leverage massive volumes of time series data to solve significant societal challenges, moving beyond mere efficiency.
Collection
[
|
...
]