I’m currently working for a company that specializes at collecting lots of of logs from its clients and running some anomalies detection tools on that data. Recently, the company has decided to add to its portfolio a security product that will be used to collect data from various sources, including logs based on packets that were analysed by the product as well as logs from security products. The product’s main goal is to automatically detect malicious activity on the network and alert when it happens. Since that I’m a huge fan of your work and have already used NuPic before for some projects I was working on, and since that the criteria for a cyber related anomaly will continuously change over time I thought that HTM would be my best bet. However, when I presented this to our chief data science officer who by the way has both B.Sc and M.Sc in neuroscience and a very strong background in machine learning and deep learning, he said that all he could find about the Numenta algorithms was written or published by Numenta itself and that he could not find any information about public ML/AI contests in which HTM was presented and evaluated against other types of deep learning techniques.
Does anyone here knows of publications and articles that were not written by Numenta that compare the HTM to other - especially cutting-edge types of deep learning techniques and/or such contests as I mentioned above?
Thanks in advance,