Machine learning applications revisited
We could approach revisiting machine learning applications in two different ways probably. First, we could do that analysis based on a very academic approach and it would be possible to just look at a Neurips Anthology visualization.[1] That type of visualization could show you exactly what people are working on in the machine learning space and you can interpret from that where current and future machine learning applications are going to be going. From a purely intellectual perspective that approach is probably going to be more interesting in terms of seeing paper clusters on topics you might not have considered. In terms of finding something novel, that is the direction to go.[2] You can filter and animate the visualization. Please go check out that link and dig into the things it shows. It is a really interesting way to look at machine learning applications. The biggest ones in the base visualization are vision, neuroscience, RL, Graph NN, classic ML algorithms, adversarial robust, classical neural networks, and optimization. I really thought games would have been a little bit bigger of a cluster, but maybe some of the game style efforts are included in other clusters.
Second, you could simply take the applied machine learning approach and make a list of what applications currently exist. That method is much more straightforward and you might already know it is something that I do as a natural part of my ongoing research. During the course of my AIOps/MLOps research I have been updating a chart of general machine learning user cases compared by scale vs. maturity. The last major update to that chart happened on July 16, 2021 based on the last updated tag in this workbook. Right now it seems like a good idea to revisit those machine learning applications here to see what seems to be changing or what new things seem to be emerging. Typically as people try to sell me different machine learning technologies, I read papers about them, or they are mentioned in a YouTube video that I watch each week. I simply add them to my list of general use cases. That list is now 29 use cases long. Sometimes somebody comes back after seeing this graphic and asks me about something that seems to be an obvious miss and those typically get added to the list posthaste.
Product recommendations
Image recognition
Video recognition
Account prioritization
Lead prioritization
Monitoring systems
Speech recognition
Marketing targeting
Fraud detection
Inventory management
General forecasting
Audio analysis
Customer experience
Voice assistants
Dynamic pricing
Deepfake detection
Email management
Sentiment analysis
Time series forecasting
Image analysis
Geospatial analysis
Streamline data
Optimize operations
General chat bots
Targeted chat bots (Service/Help)
Health check monitoring
Anomaly detection
Sound pattern matching
Sales trending
Links and thoughts:
“[ML News] DeepMind tackles Math | Microsoft does more with less | Timnit Gebru launches DAIR”
Have you ever wondered how they test a keyboard for 10,000,000 key presses? “#CES2022 ASUS TUF Gaming Laptop Durability Test 1”
“How (And Why) You Should Run Gaming Benchmarks When Buying A New GPU”
Top 4 Tweets of the week:
Footnotes:
[1] https://neuripsav.vizhub.ai/blog/ HT Yannic on sending me this direction
[2] https://neuripsav.vizhub.ai/
What’s next for The Lindahl Letter?
Week 49: Machine learning assets
Week 50: Is machine learning the new oil?
Week 51: What is scientific machine learning?
Week 52: That one with a machine learning post
Week 53: Machine learning interview questions
I’ll try to keep the what’s next list forward looking with at least five weeks of posts in planning or review. If you enjoyed reading this content, then please take a moment and share it with a friend.