{ "id": 356584, "title": "Introduction to Online Learning with resources", "url": "/competitions/tabular-playground-series-oct-2022/discussion/356584", "author_name": "", "post_date": "2022-10-01T06:42:25.921241400Z", "votes": 70, "comment_count": 33, "views": 0, "content": "

Hello all, the competition overview page refers to a method named 'online learning' and redirects to a notebook (FTLR) also. The below links are perhaps helpful to onboard onto this. Hope these help!!

\n

From my limited knowledge of online learning, I opine that this is quite different from traditional ML models. It involves generating predictions at each step, with data inflow sequentially. This is perhaps quite highly nimble and is able to adjust to evolving scenarios and model milieu. Commonplace applications of such models is in weather sciences and stock and crypto-asset forecasts where the trading and market microstructures evolve rapidly. Such models necessitate higher computational power requirements over traditional models due to their higher efficacy, this may be considered a challenge for some. Some online learning models suffer from production issues (environment drifts away largely causing a model degradation) and sometimes scalability concerns.

\n

I believe the below web locations corroborate my abridgement of these models and perhaps furnish more information in this regard-

\n
    \n
  1. https://www.qwak.com/post/online-vs-offline-machine-learning-whats-the-difference -- this highlights the A-Z of online ML
    2. \n
  2. https://github.com/online-ml/awesome-online-machine-learning -- this repo is quite useful to onboard into this area of ML
    3. \n
  3. https://www.iunera.com/kraken/fabric/simple-introduction-to-online-learning-in-machine-learning/ - this is a descriptive article corroborating my paragraph above
    4. \n
  4. https://vitalflux.com/difference-between-online-batch-learning/ -- this is another good article on the topic, mostly serves as an introduction
    5. \n
  5. https://vitalflux.com/difference-between-online-batch-learning/ - this is quite an informative introduction into this topic
    6. \n
\n

The below YouTube videos may also be useful in this regard-

\n
    \n
  1. https://www.youtube.com/watch?v=nPrhFxEuTYU
    2. \n
  2. https://www.youtube.com/watch?v=T4y25jc5NyM - this is particularly useful and introduces one to 'creme' library that is useful for online learning. This is well curated by Krish Naik
    3. \n
  3. https://www.youtube.com/watch?v=dnCzy_XKGbA - Dr. Andrew Ng sheds light on large scale models and online learning algorithms herewith.
    4. \n
\n

All the best!

", "messages": [ { "id": "1965103", "postDate": "10/01/2022 06:42:25", "content": "

Hello all, the competition overview page refers to a method named 'online learning' and redirects to a notebook (FTLR) also. The below links are perhaps helpful to onboard onto this. Hope these help!!

\n

From my limited knowledge of online learning, I opine that this is quite different from traditional ML models. It involves generating predictions at each step, with data inflow sequentially. This is perhaps quite highly nimble and is able to adjust to evolving scenarios and model milieu. Commonplace applications of such models is in weather sciences and stock and crypto-asset forecasts where the trading and market microstructures evolve rapidly. Such models necessitate higher computational power requirements over traditional models due to their higher efficacy, this may be considered a challenge for some. Some online learning models suffer from production issues (environment drifts away largely causing a model degradation) and sometimes scalability concerns.

\n

I believe the below web locations corroborate my abridgement of these models and perhaps furnish more information in this regard-

\n
    \n
  1. https://www.qwak.com/post/online-vs-offline-machine-learning-whats-the-difference -- this highlights the A-Z of online ML
    2. \n
  2. https://github.com/online-ml/awesome-online-machine-learning -- this repo is quite useful to onboard into this area of ML
    3. \n
  3. https://www.iunera.com/kraken/fabric/simple-introduction-to-online-learning-in-machine-learning/ - this is a descriptive article corroborating my paragraph above
    4. \n
  4. https://vitalflux.com/difference-between-online-batch-learning/ -- this is another good article on the topic, mostly serves as an introduction
    5. \n
  5. https://vitalflux.com/difference-between-online-batch-learning/ - this is quite an informative introduction into this topic
    6. \n
\n

The below YouTube videos may also be useful in this regard-

\n
    \n
  1. https://www.youtube.com/watch?v=nPrhFxEuTYU
    2. \n
  2. https://www.youtube.com/watch?v=T4y25jc5NyM - this is particularly useful and introduces one to 'creme' library that is useful for online learning. This is well curated by Krish Naik
    3. \n
  3. https://www.youtube.com/watch?v=dnCzy_XKGbA - Dr. Andrew Ng sheds light on large scale models and online learning algorithms herewith.
    4. \n
\n

All the best!

", "rawMarkdown": "Hello all, the competition overview page refers to a method named 'online learning' and redirects to a notebook (FTLR) also. The below links are perhaps helpful to onboard onto this. Hope these help!!\n\nFrom my limited knowledge of online learning, I opine that this is quite different from traditional ML models. It involves generating predictions at each step, with data inflow sequentially. This is perhaps quite highly nimble and is able to adjust to evolving scenarios and model milieu. Commonplace applications of such models is in weather sciences and stock and crypto-asset forecasts where the trading and market microstructures evolve rapidly. Such models necessitate higher computational power requirements over traditional models due to their higher efficacy, this may be considered a challenge for some. Some online learning models suffer from production issues (environment drifts away largely causing a model degradation) and sometimes scalability concerns. \n\nI believe the below web locations corroborate my abridgement of these models and perhaps furnish more information in this regard- \n\n1. https://www.qwak.com/post/online-vs-offline-machine-learning-whats-the-difference -- this highlights the A-Z of online ML\n2. https://github.com/online-ml/awesome-online-machine-learning -- this repo is quite useful to onboard into this area of ML\n3. https://www.iunera.com/kraken/fabric/simple-introduction-to-online-learning-in-machine-learning/ - this is a descriptive article corroborating my paragraph above\n4. https://vitalflux.com/difference-between-online-batch-learning/ -- this is another good article on the topic, mostly serves as an introduction\n5. https://vitalflux.com/difference-between-online-batch-learning/ - this is quite an informative introduction into this topic\n\nThe below YouTube videos may also be useful in this regard- \n1. https://www.youtube.com/watch?v=nPrhFxEuTYU\n2. https://www.youtube.com/watch?v=T4y25jc5NyM - this is particularly useful and introduces one to 'creme' library that is useful for online learning. This is well curated by Krish Naik\n3. https://www.youtube.com/watch?v=dnCzy_XKGbA - Dr. Andrew Ng sheds light on large scale models and online learning algorithms herewith. \n\nAll the best!", "votes": null }, { "id": "1965459", "postDate": "10/01/2022 11:24:04", "content": "

Thanks for sharing. Upvoted!

", "rawMarkdown": "Thanks for sharing. Upvoted!", "votes": null }, { "id": "1966801", "postDate": "10/02/2022 07:06:04", "content": "

Hope it helps @nitishraj

", "rawMarkdown": "Hope it helps @nitishraj", "votes": null }, { "id": "1967382", "postDate": "10/02/2022 13:25:41", "content": "

Thanks for sharing. Would you say online learning is the same as MLOps? Do you know online courses that could help on this?

", "rawMarkdown": "Thanks for sharing. Would you say online learning is the same as MLOps? Do you know online courses that could help on this?", "votes": null }, { "id": "1967803", "postDate": "10/02/2022 17:36:18", "content": "

Thanks for sharing, @ravi20076! on \"online learning\" which is quite different from traditional methods

", "rawMarkdown": "Thanks for sharing, @ravi20076! on \"online learning\" which is quite different from traditional methods", "votes": null }, { "id": "1967902", "postDate": "10/02/2022 18:43:38", "content": "

This is a new area for me too, I am also figuring it out. I shall revert shortly with some additional information for everyone's assistance.

", "rawMarkdown": "This is a new area for me too, I am also figuring it out. I shall revert shortly with some additional information for everyone's assistance.", "votes": null }, { "id": "1967905", "postDate": "10/02/2022 18:44:12", "content": "

Sure, this is a completely new area of models, totally different from traditional approaches @alvinleenh

", "rawMarkdown": "Sure, this is a completely new area of models, totally different from traditional approaches @alvinleenh", "votes": null }, { "id": "1967910", "postDate": "10/02/2022 18:46:37", "content": "

I'm happy that you shared such good things!

", "rawMarkdown": "I'm happy that you shared such good things!", "votes": null }, { "id": "1968062", "postDate": "10/02/2022 20:51:39", "content": "

thanks for sharing this helpful resources @ravi20076

", "rawMarkdown": "thanks for sharing this helpful resources @ravi20076", "votes": null }, { "id": "1968566", "postDate": "10/03/2022 06:16:18", "content": "

Thanks for sharing, @ravi20076 … this will definitely boost our knowledge

", "rawMarkdown": "Thanks for sharing, @ravi20076 ... this will definitely boost our knowledge", "votes": null }, { "id": "1968900", "postDate": "10/03/2022 09:09:02", "content": "

Happy to help!! All the best!

", "rawMarkdown": "Happy to help!! All the best!", "votes": null }, { "id": "1968906", "postDate": "10/03/2022 09:09:23", "content": "

Good luck for the assignment, hoping to learn and grow forth!!

", "rawMarkdown": "Good luck for the assignment, hoping to learn and grow forth!!", "votes": null }, { "id": "1968909", "postDate": "10/03/2022 09:09:50", "content": "

Good luck with the assignment, whatever be the final result,learning is surely going to be great!!

", "rawMarkdown": "Good luck with the assignment, whatever be the final result,learning is surely going to be great!!", "votes": null }, { "id": "1969013", "postDate": "10/03/2022 09:55:46", "content": "

Thanks @ravi20076 for sharing

", "rawMarkdown": "Thanks @ravi20076 for sharing", "votes": null }, { "id": "1969356", "postDate": "10/03/2022 12:56:07", "content": "

Thanks for the help

", "rawMarkdown": "Thanks for the help", "votes": null }, { "id": "1969560", "postDate": "10/03/2022 14:50:13", "content": "

I am happy to help @gazu468

", "rawMarkdown": "I am happy to help @gazu468", "votes": null }, { "id": "1969561", "postDate": "10/03/2022 14:50:28", "content": "

Most welcome @kinnerakiran

", "rawMarkdown": "Most welcome @kinnerakiran", "votes": null }, { "id": "1969664", "postDate": "10/03/2022 15:58:20", "content": "

Thank you sir, this is going to help me a lot 🙏

", "rawMarkdown": "Thank you sir, this is going to help me a lot 🙏", "votes": null }, { "id": "1969810", "postDate": "10/03/2022 17:42:50", "content": "

Welcome, I'll be happy if it helps @killershoaib

", "rawMarkdown": "Welcome, I'll be happy if it helps @killershoaib", "votes": null }, { "id": "1970231", "postDate": "10/04/2022 00:36:07", "content": "

Thanks for sharing this @ravi20076 It's very helful!

", "rawMarkdown": "Thanks for sharing this @ravi20076 It's very helful!", "votes": null }, { "id": "1971330", "postDate": "10/04/2022 14:31:54", "content": "

Welcome, happy to help anytime @therealoise

", "rawMarkdown": "Welcome, happy to help anytime @therealoise", "votes": null }, { "id": "1972498", "postDate": "10/05/2022 07:24:23", "content": "

Summary of this- Discussion:

\n

Online Learning vs Offline Learning(Traditional Machine learning or Batch learning)-

\n\n

After Deploying our model on the server. Let's talk about an example, In Amazon prime or Netflix when we train our model we have past year's data and in the upcoming year, their content was increased because of the new release.

\n

This can handle using two ways:

\n
    \n

  1. Pull the model from the server then merge the new data with previous data and train the model then again deploy it on the server that way traditional batch learning works. In a certain amount of time, it updates m manually.

    2. \n

  2. During deployment time we fixed some citration after a certain amount of time(could be 24 hours or 5 minutes) model collects data from the server then it trains automatically, No need to pull the model and retrain it, that's the way online learning works.

    3. \n
\n

Both have advantages and disadvantages, Based on our necessity we can choose one of them.

\n

_____Happy Kaggling____

", "rawMarkdown": "**Summary of this- Discussion:**\n\nOnline Learning vs Offline Learning(Traditional Machine learning or Batch learning)-\n- Batch-Learning works: In batch learning, We have data and train the whole data via a model. After training the model we go for production. \n\n- Online Learning works: In online learning does not need a huge amount of data, Just a certain amount of data need then train the model and deploy it. \n\nAfter Deploying our model on the server. Let's talk about an example, In Amazon prime or Netflix when we train our model we have past year's data and in the upcoming year, their content was increased because of the new release.\n\nThis can handle using two ways: \n1. Pull the model from the server then merge the new data with previous data and train the model then again deploy it on the server that way traditional batch learning works. In a certain amount of time, it updates m manually.\n\n2. During deployment time we fixed some citration after a certain amount of time(could be 24 hours or 5 minutes) model collects data from the server then it trains automatically, No need to pull the model and retrain it, that's the way online learning works. \n\nBoth have advantages and disadvantages, Based on our necessity we can choose one of them.\n\n___________Happy Kaggling__________", "votes": null }, { "id": "1972584", "postDate": "10/05/2022 08:15:41", "content": "

Great advice 👍, I have implemented in my notebook, with baseline model on train0 + online model to train on other datasets. This resolves memory limitation when handling all dataset at once.

", "rawMarkdown": "Great advice 👍, I have implemented in my [notebook](https://www.kaggle.com/code/alvinleenh/tpsoct22-ctb-online-learning), with baseline model on train0 + online model to train on other datasets. This resolves memory limitation when handling all dataset at once.", "votes": null }, { "id": "1973637", "postDate": "10/05/2022 18:35:05", "content": "

Best resources @ravi20076

", "rawMarkdown": "Best resources @ravi20076", "votes": null }, { "id": "1973951", "postDate": "10/06/2022 01:24:57", "content": "

Thanks for sharing, i think this is a great topic for everyone. This will be very helpful

", "rawMarkdown": "Thanks for sharing, i think this is a great topic for everyone. This will be very helpful", "votes": null }, { "id": "1975158", "postDate": "10/06/2022 16:16:29", "content": "

I am happy is this helps @mschuer!

", "rawMarkdown": "I am happy is this helps @mschuer!", "votes": null }, { "id": "1975161", "postDate": "10/06/2022 16:16:52", "content": "

Very befitting and comprehensive rejoinder @gazu468

", "rawMarkdown": "Very befitting and comprehensive rejoinder @gazu468", "votes": null }, { "id": "1975162", "postDate": "10/06/2022 16:17:18", "content": "

Thanks for the complement @muhammadtausif

", "rawMarkdown": "Thanks for the complement @muhammadtausif", "votes": null }, { "id": "1975195", "postDate": "10/06/2022 16:24:45", "content": "

Thanks for the Resources

", "rawMarkdown": "Thanks for the Resources", "votes": null }, { "id": "1977601", "postDate": "10/08/2022 05:46:02", "content": "

Most welcome @aayushmanjain

", "rawMarkdown": "Most welcome @aayushmanjain", "votes": null }, { "id": "2005548", "postDate": "10/27/2022 05:07:16", "content": "

Nice material, thanks for sharing @ravi20076 !

", "rawMarkdown": "Nice material, thanks for sharing @ravi20076 !", "votes": null }, { "id": "2064007", "postDate": "12/13/2022 13:07:24", "content": "

Thanks for sharing !

", "rawMarkdown": "Thanks for sharing !", "votes": null }, { "id": "2310533", "postDate": "06/20/2023 12:39:29", "content": "

Being just a beginner, these resources are going to be very helpful. Thanks for sharing!

", "rawMarkdown": "Being just a beginner, these resources are going to be very helpful. Thanks for sharing!", "votes": null }, { "id": "2458368", "postDate": "09/27/2023 14:41:15", "content": "

Thank you so much, it is really help @ravi20076

", "rawMarkdown": "Thank you so much, it is really help @ravi20076", "votes": null } ], "comments": [ { "id": 1965459, "author_name": "nitishraj", "author_url": "", "post_date": "10/01/2022 11:24:04", "content": "

Thanks for sharing. Upvoted!

", "votes": null, "replies": [ { "id": 1966801, "author_name": "ravi20076", "author_url": "", "post_date": "10/02/2022 07:06:04", "content": "

Hope it helps @nitishraj

", "votes": null, "replies": [] } ] }, { "id": 1967382, "author_name": "pastorsoto", "author_url": "", "post_date": "10/02/2022 13:25:41", "content": "

Thanks for sharing. Would you say online learning is the same as MLOps? Do you know online courses that could help on this?

", "votes": null, "replies": [ { "id": 1967902, "author_name": "ravi20076", "author_url": "", "post_date": "10/02/2022 18:43:38", "content": "

This is a new area for me too, I am also figuring it out. I shall revert shortly with some additional information for everyone's assistance.

", "votes": null, "replies": [] } ] }, { "id": 1967803, "author_name": "alvinleenh", "author_url": "", "post_date": "10/02/2022 17:36:18", "content": "

Thanks for sharing, @ravi20076! on \"online learning\" which is quite different from traditional methods

", "votes": null, "replies": [ { "id": 1967905, "author_name": "ravi20076", "author_url": "", "post_date": "10/02/2022 18:44:12", "content": "

Sure, this is a completely new area of models, totally different from traditional approaches @alvinleenh

", "votes": null, "replies": [] }, { "id": 1972584, "author_name": "alvinleenh", "author_url": "", "post_date": "10/05/2022 08:15:41", "content": "

Great advice 👍, I have implemented in my notebook, with baseline model on train0 + online model to train on other datasets. This resolves memory limitation when handling all dataset at once.

", "votes": null, "replies": [] } ] }, { "id": 1967910, "author_name": "waleedfaheem", "author_url": "", "post_date": "10/02/2022 18:46:37", "content": "

I'm happy that you shared such good things!

", "votes": null, "replies": [ { "id": 1968900, "author_name": "ravi20076", "author_url": "", "post_date": "10/03/2022 09:09:02", "content": "

Happy to help!! All the best!

", "votes": null, "replies": [] } ] }, { "id": 1968062, "author_name": "mohamedbakhet", "author_url": "", "post_date": "10/02/2022 20:51:39", "content": "

thanks for sharing this helpful resources @ravi20076

", "votes": null, "replies": [ { "id": 1968906, "author_name": "ravi20076", "author_url": "", "post_date": "10/03/2022 09:09:23", "content": "

Good luck for the assignment, hoping to learn and grow forth!!

", "votes": null, "replies": [] } ] }, { "id": 1968566, "author_name": "ankit8467", "author_url": "", "post_date": "10/03/2022 06:16:18", "content": "

Thanks for sharing, @ravi20076 … this will definitely boost our knowledge

", "votes": null, "replies": [ { "id": 1968909, "author_name": "ravi20076", "author_url": "", "post_date": "10/03/2022 09:09:50", "content": "

Good luck with the assignment, whatever be the final result,learning is surely going to be great!!

", "votes": null, "replies": [] } ] }, { "id": 1969013, "author_name": "kinnerakiran", "author_url": "", "post_date": "10/03/2022 09:55:46", "content": "

Thanks @ravi20076 for sharing

", "votes": null, "replies": [ { "id": 1969561, "author_name": "ravi20076", "author_url": "", "post_date": "10/03/2022 14:50:28", "content": "

Most welcome @kinnerakiran

", "votes": null, "replies": [] } ] }, { "id": 1969356, "author_name": "gazu468", "author_url": "", "post_date": "10/03/2022 12:56:07", "content": "

Thanks for the help

", "votes": null, "replies": [ { "id": 1969560, "author_name": "ravi20076", "author_url": "", "post_date": "10/03/2022 14:50:13", "content": "

I am happy to help @gazu468

", "votes": null, "replies": [] } ] }, { "id": 1969664, "author_name": "killershoaib", "author_url": "", "post_date": "10/03/2022 15:58:20", "content": "

Thank you sir, this is going to help me a lot 🙏

", "votes": null, "replies": [ { "id": 1969810, "author_name": "ravi20076", "author_url": "", "post_date": "10/03/2022 17:42:50", "content": "

Welcome, I'll be happy if it helps @killershoaib

", "votes": null, "replies": [] } ] }, { "id": 1970231, "author_name": "therealoise", "author_url": "", "post_date": "10/04/2022 00:36:07", "content": "

Thanks for sharing this @ravi20076 It's very helful!

", "votes": null, "replies": [ { "id": 1971330, "author_name": "ravi20076", "author_url": "", "post_date": "10/04/2022 14:31:54", "content": "

Welcome, happy to help anytime @therealoise

", "votes": null, "replies": [] } ] }, { "id": 1972498, "author_name": "gazu468", "author_url": "", "post_date": "10/05/2022 07:24:23", "content": "

Summary of this- Discussion:

\n

Online Learning vs Offline Learning(Traditional Machine learning or Batch learning)-

\n\n

After Deploying our model on the server. Let's talk about an example, In Amazon prime or Netflix when we train our model we have past year's data and in the upcoming year, their content was increased because of the new release.

\n

This can handle using two ways:

\n
    \n

  1. Pull the model from the server then merge the new data with previous data and train the model then again deploy it on the server that way traditional batch learning works. In a certain amount of time, it updates m manually.

    2. \n

  2. During deployment time we fixed some citration after a certain amount of time(could be 24 hours or 5 minutes) model collects data from the server then it trains automatically, No need to pull the model and retrain it, that's the way online learning works.

    3. \n
\n

Both have advantages and disadvantages, Based on our necessity we can choose one of them.

\n

_____Happy Kaggling____

", "votes": null, "replies": [ { "id": 1975161, "author_name": "ravi20076", "author_url": "", "post_date": "10/06/2022 16:16:52", "content": "

Very befitting and comprehensive rejoinder @gazu468

", "votes": null, "replies": [] } ] }, { "id": 1973637, "author_name": "muhammedtausif", "author_url": "", "post_date": "10/05/2022 18:35:05", "content": "

Best resources @ravi20076

", "votes": null, "replies": [ { "id": 1975162, "author_name": "ravi20076", "author_url": "", "post_date": "10/06/2022 16:17:18", "content": "

Thanks for the complement @muhammadtausif

", "votes": null, "replies": [] } ] }, { "id": 1973951, "author_name": "mschuer", "author_url": "", "post_date": "10/06/2022 01:24:57", "content": "

Thanks for sharing, i think this is a great topic for everyone. This will be very helpful

", "votes": null, "replies": [ { "id": 1975158, "author_name": "ravi20076", "author_url": "", "post_date": "10/06/2022 16:16:29", "content": "

I am happy is this helps @mschuer!

", "votes": null, "replies": [] } ] }, { "id": 1975195, "author_name": "aayushmanjain", "author_url": "", "post_date": "10/06/2022 16:24:45", "content": "

Thanks for the Resources

", "votes": null, "replies": [ { "id": 1977601, "author_name": "ravi20076", "author_url": "", "post_date": "10/08/2022 05:46:02", "content": "

Most welcome @aayushmanjain

", "votes": null, "replies": [] } ] }, { "id": 2005548, "author_name": "ashar7777", "author_url": "", "post_date": "10/27/2022 05:07:16", "content": "

Nice material, thanks for sharing @ravi20076 !

", "votes": null, "replies": [] }, { "id": 2064007, "author_name": "dwijrajhari", "author_url": "", "post_date": "12/13/2022 13:07:24", "content": "

Thanks for sharing !

", "votes": null, "replies": [] }, { "id": 2310533, "author_name": "yashiiikaa", "author_url": "", "post_date": "06/20/2023 12:39:29", "content": "

Being just a beginner, these resources are going to be very helpful. Thanks for sharing!

", "votes": null, "replies": [] }, { "id": 2458368, "author_name": "akshaykumar822", "author_url": "", "post_date": "09/27/2023 14:41:15", "content": "

Thank you so much, it is really help @ravi20076

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1965103": "Hello all, the competition overview page refers to a method named 'online learning' and redirects to a notebook (FTLR) also. The below links are perhaps helpful to onboard onto this. Hope these help!!\n\nFrom my limited knowledge of online learning, I opine that this is quite different from traditional ML models. It involves generating predictions at each step, with data inflow sequentially. This is perhaps quite highly nimble and is able to adjust to evolving scenarios and model milieu. Commonplace applications of such models is in weather sciences and stock and crypto-asset forecasts where the trading and market microstructures evolve rapidly. Such models necessitate higher computational power requirements over traditional models due to their higher efficacy, this may be considered a challenge for some. Some online learning models suffer from production issues (environment drifts away largely causing a model degradation) and sometimes scalability concerns. \n\nI believe the below web locations corroborate my abridgement of these models and perhaps furnish more information in this regard- \n\n1. https://www.qwak.com/post/online-vs-offline-machine-learning-whats-the-difference -- this highlights the A-Z of online ML\n2. https://github.com/online-ml/awesome-online-machine-learning -- this repo is quite useful to onboard into this area of ML\n3. https://www.iunera.com/kraken/fabric/simple-introduction-to-online-learning-in-machine-learning/ - this is a descriptive article corroborating my paragraph above\n4. https://vitalflux.com/difference-between-online-batch-learning/ -- this is another good article on the topic, mostly serves as an introduction\n5. https://vitalflux.com/difference-between-online-batch-learning/ - this is quite an informative introduction into this topic\n\nThe below YouTube videos may also be useful in this regard- \n1. https://www.youtube.com/watch?v=nPrhFxEuTYU\n2. https://www.youtube.com/watch?v=T4y25jc5NyM - this is particularly useful and introduces one to 'creme' library that is useful for online learning. This is well curated by Krish Naik\n3. https://www.youtube.com/watch?v=dnCzy_XKGbA - Dr. Andrew Ng sheds light on large scale models and online learning algorithms herewith. \n\nAll the best!", "1965459": "Thanks for sharing. Upvoted!", "1966801": "Hope it helps @nitishraj", "1967382": "Thanks for sharing. Would you say online learning is the same as MLOps? Do you know online courses that could help on this?", "1967803": "Thanks for sharing, @ravi20076! on \"online learning\" which is quite different from traditional methods", "1967902": "This is a new area for me too, I am also figuring it out. I shall revert shortly with some additional information for everyone's assistance.", "1967905": "Sure, this is a completely new area of models, totally different from traditional approaches @alvinleenh", "1967910": "I'm happy that you shared such good things!", "1968062": "thanks for sharing this helpful resources @ravi20076", "1968566": "Thanks for sharing, @ravi20076 ... this will definitely boost our knowledge", "1968900": "Happy to help!! All the best!", "1968906": "Good luck for the assignment, hoping to learn and grow forth!!", "1968909": "Good luck with the assignment, whatever be the final result,learning is surely going to be great!!", "1969013": "Thanks @ravi20076 for sharing", "1969356": "Thanks for the help", "1969560": "I am happy to help @gazu468", "1969561": "Most welcome @kinnerakiran", "1969664": "Thank you sir, this is going to help me a lot 🙏", "1969810": "Welcome, I'll be happy if it helps @killershoaib", "1970231": "Thanks for sharing this @ravi20076 It's very helful!", "1971330": "Welcome, happy to help anytime @therealoise", "1972498": "**Summary of this- Discussion:**\n\nOnline Learning vs Offline Learning(Traditional Machine learning or Batch learning)-\n- Batch-Learning works: In batch learning, We have data and train the whole data via a model. After training the model we go for production. \n\n- Online Learning works: In online learning does not need a huge amount of data, Just a certain amount of data need then train the model and deploy it. \n\nAfter Deploying our model on the server. Let's talk about an example, In Amazon prime or Netflix when we train our model we have past year's data and in the upcoming year, their content was increased because of the new release.\n\nThis can handle using two ways: \n1. Pull the model from the server then merge the new data with previous data and train the model then again deploy it on the server that way traditional batch learning works. In a certain amount of time, it updates m manually.\n\n2. During deployment time we fixed some citration after a certain amount of time(could be 24 hours or 5 minutes) model collects data from the server then it trains automatically, No need to pull the model and retrain it, that's the way online learning works. \n\nBoth have advantages and disadvantages, Based on our necessity we can choose one of them.\n\n___________Happy Kaggling__________", "1972584": "Great advice 👍, I have implemented in my [notebook](https://www.kaggle.com/code/alvinleenh/tpsoct22-ctb-online-learning), with baseline model on train0 + online model to train on other datasets. This resolves memory limitation when handling all dataset at once.", "1973637": "Best resources @ravi20076", "1973951": "Thanks for sharing, i think this is a great topic for everyone. This will be very helpful", "1975158": "I am happy is this helps @mschuer!", "1975161": "Very befitting and comprehensive rejoinder @gazu468", "1975162": "Thanks for the complement @muhammadtausif", "1975195": "Thanks for the Resources", "1977601": "Most welcome @aayushmanjain", "2005548": "Nice material, thanks for sharing @ravi20076 !", "2064007": "Thanks for sharing !", "2310533": "Being just a beginner, these resources are going to be very helpful. Thanks for sharing!", "2458368": "Thank you so much, it is really help @ravi20076" }, "source": "meta" }