Recently, I completed 'Machine Learning Onramp' course from Mathworks. The course was good enough with illustrations. However, after completing it, I am still not sure about the application of Artificial Intelligence using MATLAB software. How to proceed further keeping in mind that there are still powerful tools such as 'RStudio' in the market? Any suggestions??
Dear researcher
as a Matlab user for years, i can tell you that Matlab platform is not real suitable in a such application, Python is the best option
I would also recommend starting with Python by using the scikit-learn package to do multivariate linear regression, logistic regression, SVM, and random forest models. Once you go into neural networks and deep learning you can learn about the PyTorch, Keras and Tensorflow packages next.
learn python. but for starters do not go for scikit-learn keras or tensorflow. try to instead build from scratch using numpy and pandas. this will provide you with more intuition towards how actually these algorithms work.
Dear Ravi Verma,
We can do something with MATLAB in AI or Machine Learning: AI models in MATLAB can be deployed on embedded devices or boards, edge devices in the field, enterprise systems, or the cloud. We can harness the power of Machine Learning to use data to make better decisions. MATLAB makes machine learning easy. With tools and functions for handling big data, as well as apps to make machine learning accessible, MATLAB is an ideal environment for applying machine learning to your data analytics.
Indeed as the colleague points out there is this challenge between R and Python, however we can say that RStudio’s recommended professional data science solution for every team.
For more details and information about this subject I suggest you to see links and attached file on topic.
https://www.mathworks.com/campaigns/offers/ai-with-matlab.html
https://www.mathworks.com/solutions/machine-learning.html
https://moderndive.com/1-getting-started.html
https://learningstatisticswithr.com/book/hypothesistesting.html
https://rstudio.com/
Best regards
First, you have to choose an aspect of AI to look into. You could choose Evolutionary Algorithm, understand how it works and the kind of problems it can solve. Then get a problem, define your approaches, strategies and models as the case maybe and evaluate using Matlab. Since you just finished learning Matlab, I suggest you restrict to using Matlab for now, to enable you master it and acquire some insights on how these algorithms work. Thereafter, you progress to using Julia. Julia performs better than Matlab and even Python. I just concluded a project on Al applied on dependable systems using an evolutionary algorithm adopted in Matlab and Julia.
Ravi Verma , Python allows you more flexibility to write several lines of code and need not worry about run-time relative to MATLAB. You can create several exciting functions() in Python. You can do this in MATLAB too. But I found it a bit slow when running several codes at a time. Although Prof. Andrew Ng's course on ML allows you to learn it via GNU Octave/MATLAB, it can however be a start. But, if that is your career, I would suggest you to go for Python because it is open-source unlike MATLAB, allows more flexibility, and probably a bit faster than MATLAB. Another reason why I am not recommending MATLAB is that it requires a license and its confirmation every time you update it. So, what it means is if you go past the expiration date, you may no longer be able to access and perform coding in MATLAB until you renew it, which is inconvenient for many people. You don't need to worry about such things in Python.
Based on my own experience first, you need to be familiar enough with the basic algorithm and have idea in your mind how you are going to approach your case using these algorithms. Regardless, It really does not matter if you make use of Matlab or other software, for Matlab toolbox in AI broadly for different applications, particularly built-in toolboxes in Matlab facilitate your intention a lot.
Obviously Python has the best libraries to preform any machine learning algorithm. In my opinion, it is quite easier than MATLAB to learn. On the other hand, MATLAB has been developed for a lot of scientific approaches since many years ago and it has became a mature tool. Many engineering software has the option to be coupled with MATLAB. In many cases, you have to use a primary model to develop something other than a machine learning model. Real-time data transfer to Python still is a problem that should be addressed in the following years.
Mehrdad Vojdani Agreed. Python, being an open source can provide more accessibility and opportunities in solving AI problems.
I would say that MATLAB has limited features while Python has largest set of features. Also one cannot optimize for speed on a particular hardware using MATLAB which seems to be a critical requirement form ML and AI community. Julia programming a new language which can be advantageous for both development and deployment for computationally intensive problem and it also removes two-language problem since you write a deploy every part of code exclusively in Julia.
Python has always been the best since its creation because of its simplicity.
- Badges
- Science topic