Personalized search is the way of the future; this post will cover the most popular #MachineLearning features and techniques to get the job done👇
https://t.co/8ZpowvYfw3
Wondering what goes into #FeatureEngineering for Fraud Detection? Learn all about it in our latest post by @nickparsons on the @FennelAI blog 🔗👇
https://t.co/yhWgdoWc2O
Next ML meetup in SF/Bay will feature topics on real-time ML. @nikhilgarg28 from @FennelAI ,@Igor__Markov from @MetaAI will dive deep into real-time ML, architecture & e2e ML platform, with 🍕🍺🎁 @nickparsons
⏰ Dec 13, In-person | Virtual
📌 https://t.co/3gPIWCMIFf
Learn the 5 serving stages of #RecSys 👇
🔍 Retrieval of relevant items
⏳ Filtering to narrow down options
🏗️ Feature extraction to capture key characteristics
⚽️ Scoring to evaluate relevance & quality
🥇 Ranking to present top choices to users
https://t.co/kX4tuQNvVu
Once ANY marketplace hits enough scale to meet product market fit, THE most powerful growth lever almost always is improving "matching" via ranking/recsys.
It turns out realtime #ML models cheat on tests, too 👀
Learn about the two most common types of information leakage and how to eliminate them 👇
https://t.co/3dz60A6oSv
Introducing our second post on #FeatureEngineering for #RecSys. Click the 🔗 below to learn about:
🕹️ Max/min pool-related options
🧮 Wilson intervals to normalize rate features
🪟 Rolling time windows
🔌 Making counter/rate features realtime
https://t.co/kOotlBiQkU
A lot has been speculated about TikTok's recommendations. This is the first paper I've read by the team, and it has many interesting details: expirable embeddings, parameter server, online training... Good #recsys stuff https://t.co/r8UBCrnZi8