Anomaly Detection Benchmarks
The conversation delves into the challenges of establishing benchmarks for streaming temporal anomaly detection, highlighting a lack of standards compared to spatial data. By creating the Numenta anomaly benchmark, the team aims to facilitate comparisons among various models, including LSTM and open-source projects from Twitter and Etsy. They discuss the significance of curated data sets, such as taxi call records from New York City, which reveal identifiable anomalies tied to real-world events.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
The Biological Path Towards Strong AI with Matthew Taylor - #71
Related Questions