1. Image processing model for Medical sector
2.Natural Language Processing
3. Speech to text conversion
4. Deep learning model for disease diagnose
5. Deep Learning model for Image processing in Nanoscience & technology
Hi
I suggest the following:
- Neural networks
- Word embeddings / distributed representations
- Stochastic gradient descent / convex optimization
- Hyperparameter optimization
- Crowdsourcing
Capsule Network is worth exploring. i am currently working on this.
Thanks,
Pankaj
Hi Ahuja,
I am a PhD in machine learning-based control, so I I suggest the following:
- Reinforcement learning
- Adaptive dynamic programming
I hope you choose the right direction .
I would suggest you to explore how swarm intelligence can be used in deep learning. I am sure you will be amazed with the potentials and findings.
I would suggest Extreme Learning Machine (ELM). It is a new version of machine learning that is extremely fast compared to other ML techniques. Refer to : 'Extreme Learning Machine: Theory and Applications', by Guang-Bin Huang, Qin-Yu Zhu, Chee-Kheong Siew.
If you are interested in Machine and Deep Learning, make sure whatever subject you choose there is enough training data. Anomaly detection in medical imaging is certainly an interesting research subject but it might be difficult to get enough training data. e. g. Twitter has a streaming API, opinion mining using Twitter data might be a subject and there is plenty of data available.
I would suggest to do research on unsupervised machine learning applications. Currently, unsupervised learning is one of the trending study topic and as of now, very little part of it explored by scientists and researchers.
Following up on my suggestion for research on Extreme Learning Machine (ELM), here is a paper that would be highly useful to get a jump start: 'High-Performance Extreme Learning Machines: A Complete Toolbox for Big Data Application", by Anton Akusok, Kaj-Mikael Bjork, Yoan Miche, and Amaury Lendasse; IEEE Access, July 17, 2015. Not only do they tell you where to get the software (free, I think, since its Python), but also discuss common hardware (PC's, GPU's, etc) , Small Data sets and Large Data sets (available from University of California, Irvine 's Machine Learning Repository). Results from ELM are compared with other machine learning techniques. I am personally interested in Higgs Boson and that's a 1 million samples dataset. Suggested research: development of multi-layer ELM; transfer learning techniques.
Cognitive Computing will be most promising topic with respect to your interest
- Badges
- Science topic