2. Cyber Network Data Processing; AI Data Architecture

MIT RES.LL-005 Mathematics of Big Data and Machine Learning, IAP 2020
Instructor: Jeremy Kepner, Vijay Gadepally

This lecture started with anomaly detection and data conditioning in machine learning and later continued to explore the AI data architecture in data processing.

License: Creative Commons BY-NC-SA

[douglashurd]

I didn't know that there was anyone else who was so excited about naming files and folders. I no longer feel alone in the universe!

[swavekbu]

35:03 - this pipeline represented is just like many terms in computer science, glorified terminology for concepts that have existed forever in stats and math and data analysis. Raw, parse, ingest, query, analyze . . . in other words, just getting your data file together and knowing what you're analyzing, that's pretty much it. No need for a "pipeline" term or fancy words to describe a process that has existed forever. "Ingest" data? OMG.

[muhsinkhalif]

Wonderful lecture. Thanks for making it available on youtube

[lazywarrior]

@21:51 Just curious how the "no attack" cases are tagged in this study. Is it the traffic data downloaded from that JP site? If so, then this is just the detection prob of the fake traffic that you inserted, right? You may also insert some normal traffic using that app, but then there's no point in using that JP data. I am confused here.

[pnemonic]

I am surprised that at #50:55 they don't use MM/DD/YYYY and MM-DD-YYYY naming format

[archibaldgoldking]

its something new, what happens here, i like it very much, because i know mathematics, and like computer sciences

[josephbsmalljr]

Please fix the audio and repost. Background music(?) is too loud relative to speakers' voices.