Shaikh Quader

Lead AI Architect, IBM Db2 Database Products

[Paul Zikopoulos @BigData_paulz]

Lots of #AI Foundation Models talk these days (the AI behind #ChatGPT). Did you know that @IBM used its https://t.co/UqGnM26mY0 platform to build 20+ of these models that are safe data trained across a myriad of domains. IMHO this is where the future of AI will go: +AI > AI+.

[JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱 @JFPuget]

Worth a read.

[Christoph Molnar 🦋 christophmolnar.bsky.social @ChristophMolnar]

Supervised machine learning models are deployed everywhere.

It's an open secret that all models have a huge problem:

Performative prediction - when predictions change future outcomes.

How to spot and handle this problem: A thread 🧵

[JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱 @JFPuget]

The last filter (acceptance) is the hardest to overcome. I faced it in ML in other domains than medical, and I have faced it in operations research (OR).

The solution is to work on a collaborative system where AI or OR models help people, and not try to replace people by models.

[Sebastian Raschka @rasbt]

@JFPuget @giffmana @ducha_aiki 💯! Related anecdote: 2 years ago, I gave a talk at the ML-focused seminar series at our computer science department. I mentioned that XGBoost is nowadays the way to go for tabular data. A senior CS faculty (who publishes at NeurIPS etc.) asked me what XGBoost was 🤷‍♂️

[Shaikh Quader @ShaikhQuader]

https://t.co/e18MsfDnM0

[Shaikh Quader @ShaikhQuader]

Tomorrow, May 28th, University of Toronto Machine Intelligence Student Team (UTMIST) will showcase several student projects where ML was applied to solve practical problems from different fields, such as medical and financial domains. I will be one of the…https://t.co/jlp1QLaA4i

[Shaikh Quader @ShaikhQuader]

https://t.co/zmvcGNwa65

[Shaikh Quader @ShaikhQuader]

9°C outside - the warmest after a few months of cold winter. The sun is shining, and the snow is melting. I didn't miss the opportunity to go outside this morning. Completed a 4.5km outdoor run/walk, burned 421 calories, and decluttered my mind. Recharged…https://t.co/Yjg6GFMnhZ

[Shaikh Quader @ShaikhQuader]

Nearly all my classification projects had two challenges in common: (1) many examples were missing class labels and (2) many existing class labels were incorrect. In the following session, I will talk about our framework that can address both challenges.…https://t.co/pCpY9mVmus

[Shaikh Quader @ShaikhQuader]

When searching for excellence, before comparing with tops in my field, perhaps, a more pragmatic and profitable exercise is to compare myself from today with myself from yesterday? Did I improve, slide, or remain the same? Tiny improvement every day compo…https://t.co/amnhjn4peC

[Shaikh Quader @ShaikhQuader]

https://t.co/2cifC3UgET

[Shaikh Quader @ShaikhQuader]

I feel, often it's worth putting a pause before learning something new. Rather, I should reflect on what I already learned and how much of that learning I had put into practice. After acquiring a skill or a piece of knowledge, if it's not applied soon eno…https://t.co/VwuG2pPgFh

[Shaikh Quader @ShaikhQuader]

Learning and creating are two wings of a knowledge worker. Without these wings that can't fly well. I am glad that I could continue learning new tech and business skills in 2021 and achieved this learning recognition at IBM.

#learning https://t.co/xk0KHr4GGL

[Shaikh Quader @ShaikhQuader]

https://t.co/cPXNvvIpHt https://t.co/WtpJZiO09w

[Shaikh Quader @ShaikhQuader]

5 years ago, when I was a newbie to ML, I spent long hours in manually importing ML datasets from their databases to my python notebooks.

Then I spent several days, sometimes a few weeks, in preparing the data and finding the bes…https://t.co/CaAVqzMOEe https://t.co/KdyIl6Ht70

[Shaikh Quader @ShaikhQuader]

https://t.co/cUnavgoVKm

[Shaikh Quader @ShaikhQuader]

Next week, CASCON x EVOKE conference will take place virtually.

I will deliver 4 session - an industry talk on AutoML with IBM Db2, a half-day hands-on workshop on ML with Db2, and two expos.

register for free at https://t.co/eyykH14S89.
#CXEtechco…https://t.co/MiIgY02BVQ

[Shaikh Quader @ShaikhQuader]

Earlier, whenever I had a writing work, I opened a blank document in Microsoft word and started writing. Often, after producing a few lines, I paused and started fixing what I just wrote. It took me hours to even finish a few pages of written piece. I dre…https://t.co/8je90tp2dd

[Shaikh Quader @ShaikhQuader]

#mlmodels are both software and data. Similar to #cicd for software, production-grade machine models also need to go through #cicd steps:
CI for ML = automated training and testing of the #mlmodels
CD for ML = automated deployment of the new models

Ma…https://t.co/f703cNUkSx