Anomaly Detection within Machine Learning on Logs
GPTs (Generative Pretrained Transformers) based on Large Language Models are great for a lot of challenges. But they're not trained to find outliers within your log data.
In this brief, informative, and useful session, Christopher Crowley will discuss the concept of a variational autoencoder, then show how you could implement this to train an autoencoder based on your logs.
After training on your log information, you would then implement the concept to look for outliers within your log data, to surface weird things to analysts for review.
There will be about 20 minutes of theory, and about 30 minutes of practical demonstration using a jupyterlab notebook, python, and tensorflow.
Even if you're not a programmer, this will be a worthwhile session to understand what's possible with machine learning and your own log information.
