( This is a placeholder for my learnings in the context of MLExpert , updated almost daily, until I complete the course)
Foundational Knowledge in Machine Learning
Supervised Learning
When machine is learning from the provided data, it is supervised, in the sense that, for it to make sense of data, some labeling information of the data is provided to it, so that, it can map labels with the data.
For example: label all images of cats with 'Cat' / label all images of dogs with 'Dog'
Unsupervised Learning
In this also, machine is provided with data, but it is not supervised, in the sense, that, there is no hand holding here. Somehow, machine has to figure out grouping of data and has to create cluster of data points having some sort of similarity.
For example: give a stream of tweets from twitter and let the machine try to find patterns from the textual data. It would clusterize the tweets according to the category of topic associated with the tweet. It could be related to sports or tech or politics etc
Deep learning
In this also, data is provided to the machine, but the data would not be clearly structured, for example: image data. This is where neural networks are involved, specifically they are set of functions which try to determine pattern from data. There are lot of hidden layers which try to decode the pattern associated with the data.
Recommendation Systems
In this, for the end user, something needs to be recommended. For example, in case of ecommerce platform, we need to recommend products which could be potentially bought by the end user. Or, in case of Youtube / NetFlix, we need to recommend videos, which could be potentially watched / viewed by the end user. ( more time you spend on youtube, its an indicator to advertisers that, the audience is engaged with this platform and it makes sense for the advertisers to spend on Youtube. more time you spend on netflix, its an indicator that, Netflix is providing content suited to your taste, so, the end user is maintained with the perpetual subscription. ) So, ultimately, there are bunch of videos/ items to be recommended to user. Recommendation algorithm picks the best video to be shown to the user
Ranking
In this, for the end user, some kind of ranking to be done on content which is being recommended to the end user. So, what's the first video, what is the next best video to engage the user, what video after that, so, there is already set of videos which is stored/created. Ranking algorithms picks the best set of videos and ranks them and the videos are recommended in the order of rank of that video.
Top comments (0)