This article is for you, if you’re in the final year or pre-final year of your college, or you are a software engineer and you’re trying to move to the machine learning domain, then this article is for you.
So, you may be thinking, what machine learning projects should I have so that I get shortlisted for machine learning engineer jobs.
And just to give a brief background, there are multiple machine learning jobs. So some of them, you can see like, these are like research scientists, data scientists, and engineering jobs like ml engineer or deep learning engineer.
This article is intended for entry-level Machine Learning Engineering rules, or data science rules both entry-level.
So if you’re already a software developer, then this would be much easier for you, you have a good programming background, and you will be expected to do programming in the role of machine learning engineering as well.
The only thing you need to make sure that you are able to show your interest in this domain. And you have done some projects which are significant.
So I have picked the top 10 of these projects. And one thing I would suggest is that don’t just focus on the projects that I’m going to mention But also try to derive something out of that.
And try to create your own application from that, what extra you can add on those because many of these projects would be available in public GitHub repository also.
So you need to build an understanding of that. And also, you need to add something to that, so that you have done some extra on top of that.
1. Sentiment Analysis
If you want to work in the computer vision domain, Then you focus more on computer vision projects, which involve convolutional neural networks mainly, So here, you should search for the projects online, you will get many, RG projects.
But you should filter out those because there will be many people who would be doing someone like you also doing, they may have been doing from for some quality projects.
So next, you think about how to get the data set for this. So you can look into those papers. So all the papers mentioned what data said they’re used.
So you can search for the keyword data set in the paper, and you will find multiple occurrences of these.
Many of those data sets would be available in the open source in the public domain. You can download those some of them may not be available, you may have to request the organization that created those data sets.
So the first task will be to get those data sets. And most of these resources. These are projects they also share the GitHub repo.
So if some paper is popular, then multiple people implement those papers. So you will find multiple GitHub repos for the same projects.
So your first step would be that you should try to run replicate those on your system. So, get the data set, get their implementation, and see if you can replicate it on your system or not.
So, once done, you should understand what are within those projects, you should go through the paper and see what model they have used, what layers they have used, and why they have used,
So try to go through that paper and build your understanding. And slowly you will find reading papers much easier than you do in the first few papers, you may find some difficulty, but slowly, you will build that habit.
So I would highly recommend doing this sentiment analysis and also pick a framework of your choice.
2. Recommendation System
You see such things is happening all around on Netflix, you open Netflix, you will see some recommendations based on your what movies you have seen in the past, and so on.
And in Amazon Prime or even some movie booking apps, what movies you have booked in the past, they will recommend based on your interest and this cannot be just movie recommendation, similar recommendation can be applied in other fields.
So, this is a this has a high business value and multiple companies would prefer someone who has a good experience in recommendation system building recommendation systems.
So, first to start with a movie recommendation system you will find multiple implementations of these then you should think about what extra you can add.
So, you can think about what other domain I can use this recommendation system where it has not been used that much. So you can try something like that.
3. Face Recognition
So you may see that you upload multiple photos on some social platform, and it’s your tag one of the faces like this person is the xxx, then that will automatically try to suggest the same name if it finds a similar face, So, this is clustering.
So, try to explore all of these and try to build some application out of it, And you can write that in your region.
So, build some chatbot, and then any site you visit most of the time, first, they will direct you to some automated chat bots, which can answer your questions based on some standard queries.
And if you’re not satisfied, then only you contact the customer care for one to one, chat with some person to the chat bots are really getting popular.
So try to implement one try to learn how to develop chat bots.
4. Neural Style Transfer
it’s one of the recent works maybe after 212 thousand 1516. These got popular after Ganz became popular, generative adversarial networks. So these are all derivatives of those. So this is called neural style transfer.
So the main concept in style transfer is that every image has two components. One is style. And the other one is content.
To give you a small example, let’s say there is a photograph of yourself, click to buy some mobile, then another photograph in black and white, then another sketch, let’s say somebody has killed your photograph.
So if you saw this photograph to somebody, some of your friends, they will recognize all the three photographs.
And they’ll say that in all of those photographs, you are there. So they are recognizing the content. So they have the same content, although the style are different.
So you can represent the same content in multiple styles. For example, we have, let’s say, text, a paragraph. And I have written that paragraph in my own handwriting somebody has typed it in some italics, a times Roman font, somebody has written in his or her own handwriting.
So all are writing the same paragraph. So the content is same, but style is different. So that’s the concept here in style transfer.
5. Dress/Makeup transfer
Again, it may have some business value, because let’s say you are browsing through some dress for yourself on Amazon or some other eCommerce platform.
So, they would like to transfer the dress to a common person a common model. And so you that model wearing that dress, whatever dress you pick.
So here, for example, you can see this is the user, And this is the reference image, whose dress we want to transfer to this. And here that dress has been transferred to this user.
Similarly, it can be used in makeup transfer. So this is let’s say, that actually means of, of the person.
And this is the makeup image that we want to use on this person, So pretty similar to this concept, transferring just transferring makeup.