LexSiga

9 ms·

61.
AI will destroy us all. Animal edition(twitter.com)
1 points by LexSiga 4y ago|0 comments
62.
All feature types in Machine learning(hopsworks.ai)
2 points by LexSiga 4y ago|0 comments
63.
Feature Stores vs. Data Warehouse(hopsworks.ai)
3 points by LexSiga 4y ago|0 comments
64.
Dont worry about bias in AI, says Deloitte(protocol.com)
2 points by LexSiga 4y ago|0 comments
65.
Machine Learning; Testing Features and Pipelines in Python(hopsworks.ai)
1 points by LexSiga 4y ago|0 comments
66.
by LexSiga 4y ago
While this article is on offline testing with pytest we are also preparing a second one on continuous integration tests.
67.
Google gets on the feature store bandwagon(cloud.google.com)
1 points by LexSiga 5y ago|0 comments
68.
by LexSiga 6y ago
Well, all good and nice, but we are so far behind on data infrastructure, development, and investment that I do not know how realistic it is to set those targets when the basics are not here.
69.
Detecting financial fraud using GANs with Nvidia(developer.nvidia.com)
18 points by LexSiga 6y ago|0 comments
70.
AI/ML needs a key-value store, and Redis is not up to it(logicalclocks.com)
13 points by LexSiga 6y ago|0 comments
71.
by LexSiga 6y ago
ok.
72.
by LexSiga 6y ago
I am absolutely convinced by the argument around the needs and such, much less about the imminent collapse, Fraud and unreliability of the digital ads is not new, it is not even a bug, it is a feature of the people who work in advertising;
73.
by LexSiga 6y ago
That is arguable, at best it is functional enough so that people can use it, at worst it is not functional for squat and it is mostly dominant because of the incubator attached to it. In fact I wonder the rate of drop vs new user retention
74.
by LexSiga 6y ago
I did not realized that aesthetic was not correlated with function. Probably a clear visual understanding to the actual value of something is meaningless, I suppose.
75.
by LexSiga 6y ago
I love that this gets some love. I found it odd that the website is in itself not very pretty (like... that kind of goes against the main selling point of the community).
76.
by LexSiga 6y ago
That is a bit disingenuous; it maybe has issues in syncing with folders or such, but honestly when you work directly in the drive for your document creation and sharing and co-authoring, its a breeze. It is good for certain use cases, and
77.
We created a community for Feature Store resources(featurestore.org)
5 points by LexSiga 6y ago|0 comments
78.
by LexSiga 6y ago
You are not wrong on the buzzword, but you are not absolutely right if you suggest that it is merely that; this is a recurring question and interrogation in that specific area. The fact that it happens to be kinda buzz worthy is a collatera
79.
One function is all you need for ML Experiments(logicalclocks.com)
10 points by LexSiga 6y ago|1 comments
80.
by LexSiga 6y ago
Some sad soul just lost their job. Also thanks; I fixed it.
81.
by LexSiga 6y ago
As this topic will inevitably become more trendy find some some additional interesting resources on the subject as well: - https://www.quora.com/What-are-the-implementation-challenges... - http://featurestore.org
82.
How to build your own feature store for ML(logicalclocks.com)
114 points by LexSiga 6y ago|23 comments