Final Three Pitches

with No Comments
  1. Machine learning to automatized music composition

This project aims to build a system that uses machine learning algorithms to generate original music compositions based on original ones. This project will be studying machine learning techniques for pattern recognition. 

The feedback I received from this seems to be more approachable than my first pitch, but I haven’t been able to receive more feedback because nobody has experience with this matter. 

However, Doug questioned if there are Machine learning packages that I could use to build my project. Because in my project I aim to produce a new product from already existing ones. So, do we have packages to generate a new thing? It is possible to configure a train to produce something new. 

I am currently getting in contact with the music department because there is a professor who teaches a class, “Making Music with Computer.” 

I learned about software for music composition called “Max” which is a visual programming language for music and multimedia. I also learn about authors as David Cope, who he has a lot of work regarding artificial intelligence in music. I also investigated different types of music algorithms for composition and am currently looking for a specific problem or field to explore in this area. 

Figure out because off all the complexity of this keep things simple.

  • translational models
  • mathematical models
  • knowledge-based systems
  • grammars
  • optimization approaches
  • evolutionary methods
  • systems which learn
  • hybrid systems
  1. Detecting non-child-appropriate content in videos. 

Nowadays, it is very easy to get access to a lot of content online, including music, articles, games, videos, and movies. In the last couple of years, the use of video platforms such as YouTube has become very important for education, entertainment, hobbies, etc. However, YouTube is a platform where anyone can upload content, and not all content is appropriate to everyone. It can be because they may include sensitive content or bad words. This project will aim to use deep learning and machine learning for video analysis. It will require image processing to process video clips. It can also work with pattern recognition. This project aims to detect images and audio from the videos. 

Challenge: video processing, 

Data set of video file tag if they are not appropriate for children. 

Potential dataset: (Restricted) (adult content detention)

Youtube Data API (meta data)

Good Resources
  1. Real-time sign language translator to text.

This project aims to use machine learning to accurately translate real-time sign language to text by capturing sign language gestures using a camera (more likely from a computer’s web camera) to text. This system should be able to recognize various signs. I would like to study computer vision techniques. 

From the input I received from Doug, this is a very ambitious project. We thought about what would be more convenient. Translate from moves gestures to text or from text to moves gestures. The first option seems to be more complicated because computer vision techniques can be a vast area. More than understanding computer vision techniques, but will also be required to recognize sign language and the context. To be able to train machine learning, I will require a lot of data that may not be easy to obtain. 

This project will aim to have a friendly user interface. So it will be easy for the user to navigate through the translator.
  1. Machine leaning for the prediction of stroke diseases

Stroke is a cerebrovascular disease and is a significant causes of death. It causes significant health and financial burdens for the individual and the system. There are many machine-learning models built to predict the risk of stroke or to automatically diagnose stroke using predictors such as life factors or images. However, there has not been an algorithm that can predict using lab tests.

Leave a Reply