Talk:Machine learning

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

neuromorphic computing[edit]

someone should really include neuromorphic a discusion on memristors and thair connection to neural networks would also be nice

Wrong date[edit]

The date when Arthur Samuel wrote the first program is 1952 as wikipedia and many other online websites say, however this article says 1959, which should be corrected.

Approaches[edit]

Training models[edit]

  • Should optimization algorithms (in general) be discussed on this Wikipedia page? If so, this is likely the best place to do so.
  • There is currently no mention of fully decentralized machine learning methods, such as CHOCO-SGD.[1][2] This should definitely be fixed!

References

  1. ^ Kairouz, Peter; McMahan, H. Brendan; et al. (10 December 2019). "Advances and Open Problems in Federated Learning". arXiv preprint. Retrieved 20 November 2020. {{cite journal}}: Explicit use of et al. in: |last3= (help)
  2. ^ Koloskova, Anastasia; Stich, Sebastian U.; Jaggi, Martin (1 February 2019). "Decentralized Stochastic Optimization and Gossip Algorithms with Compressed Communication". arXiv preprint. Retrieved 20 November 2020.

Material brought here from artificial intelligence[edit]

I added a few paragraphs that had been cut from artificial intelligence (which was too long). They now appear in Machine learning § Overview and in Machine learning § Limitations. Feel free to improve the merge if you think that's needed.

Also, if you're an expert, take a look at Machine learning § Other limitations. I'm not an expert, and I don't know if these should have their own header.

One last thing: I don't know whether the contribution in the box below is a valuable contribution or not, and I don't know where you all might want to put it in this article, so please help me find a place for it, or let me know if we should throw it away. ---- CharlesGillingham (talk) 02:47, 13 October 2021 (UTC)Reply[reply]

Learners can also work on the basis of "Occam's razor": The simplest theory that explains the data is the likeliest. Therefore, according to Occam's razor principle, a learner must be designed such that it prefers simpler theories to complex theories, except in cases where the complex theory is proven substantially better.[1]

Machine learning is also present in modern day mining[edit]

Machine learning and deep learning have increasingly attracted interest over the last five years and we often see these terms applied in the context of mineral exploration, mine exploitation and geoscience studies.

I recommend adding mining as another Application to machine learning. In the mining industry, both recent start-up companies and well-established mining and service companies are implementing machine learning in all facets of their work. A structural geologist at the mining company I work at, SRK Consulting, wrote a published paper on this and presented it at the Geological Association of Canada in 2019.

[2]

ML as prediction based on passive observations vs AI as active agent[edit]

I don't see how making the following assertion makes any sense:

"The difference between ML and AI is frequently misunderstood. ML learns and predicts based on passive observations, whereas AI implies an agent interacting with the environment to learn and take actions that maximize its chance of successfully achieving its goals.[26]" in section "History and relationships to other fields", subsection "Artifical Intelligence"

Two points: (1) The second part of that sentence describes reinforcement learning, which is one of the three ML paradigms. (2) The idea of intelligent agents only emerged in the 1990s, but the field of AI existed long before. Not all of AI is focused on agents interacting with an environment.

I personally think that this assertion is plain wrong and at the very least extremely confusing to people who are not already familiar with the field.

Both terms are wide-ranging and variable. The statement that you are noting is basically an author putting forth their opinion in that area, and in a way that claims it is the only opinion / answer. IMO it should get deleted or given attribution type wording. North8000 (talk) 19:48, 25 January 2022 (UTC)Reply[reply]

Semi-protected edit request on 3 September 2022[edit]

Examples of Machine Learning The Examples of Machine Learning are:-

Image and Speech Recognition: These are one of the most common uses of ML. Image Recognition is the ability of software to identify objects, places, people, writing and actions in images. Speech Recognition is the ability to translate spoken words into the text. The common goal of image recognition is to classify detected objects into different categories. It is also known as object recognition. Speech recognition focuses on the translation of speech from a verbal format to a text one whereas voice recognition just seeks to identify an individual user’s voice.

Medical Diagnosis: Machine Learning can detect patterns of certain diseases within patient’s electronic healthcare records and inform clinicians of any anomalies. By its developing algorithms it provides information to the machine that can help in imaging and analyze human bodies for abnormalities. Hence, Machine Learning is making healthcare smarter.


Prediction: With the help of Machine Learning, GPS navigation predicts traffic ratio through central traffic managing servers. Businesses use ML in order to recognize patterns and then make predictions about what will appeal to customers and help make a better product.


Finance: Data scientists are always working on training systems to detect flags such as money laundering techniques, which can be prevented by financial monitoring, enhanced with ML. Algorithms of Machine Learning can be used to detect transactional frauds by analyzing millions of data points that humans might miss. Anushka544444 (talk) 07:07, 3 September 2022 (UTC) Introduction to Machine LearningReply[reply]

Objected – direct WP:COPYVIO, low quality, likely even spam. --Zac67 (talk) 07:16, 3 September 2022 (UTC)Reply[reply]