This curriculum consists of 4 concise mandatory semesters/parts and one optimal segment; an Introduction to Artificial General Intelligence.

1. The first semester, concerning the python programming language (and regularly used machine learning libraries in python).
2. The second semester, concerning using github. (Github will contain your work, and employers prefer to see healthy github pages!)
3. The third semester, concerning the main topic, Deep Learning!
4. The last and fourth semester, concerning applying knowledge from the prior semesters, and attempting to solve a problem faced by preferably the Jamaican country or elsewhere. (The code/idea developed is yours, and is not the property of MLJI.)

1. Watch and do python lessons found in this fun youtube video playlist, from lessons 1 to 32. Remeber, out of 56, you'll do only the first 32. (Don't worry, because most of the videos in this fun playlist are roughly 5 minutes long!)

2. You're done with the First Semester, Part A. Now onto the First Semester, Part B!

The author of the playlist above is Bucky Roberts.

1. Do the lessons found on this page.

2. You're done with the First Semester, Part B. Now onto the First Semester, Part C!

The author of the tutorial above is Stanford. (The tutorial above was suggested by this page.)

1. Do the lessons found on this page.

2. You're done with the First Semester, Part C. Now onto the second semester!

More python pickle information can be found on the Pickle Library Page.

1. Do the tutorial found on this page.

2. You're done with the Second Semester. Now onto the third semester!

The author of the tutorial above is Roger Dudler.

1. Follow closely, this 9 step neural network construction tutorial by Eden Au.

2. Intuitive understanding of basic machine learning components, such as fundamental artificial neural network structures, is important for maintaining the desire to continue to pursue the field of artificial intelligence/machine learning/data science. If you still desire a more intuitive description of the basic artificial neural networks described in item 1 above, read the text descriptions and watch the video content found in my book "Artificial Neural Networks for Kids". (Get the book from this Amazon url, or get it free from this Research-Gate url).

3. As a pre-requisite only for item (4) below, if not reasonably familiar with Java, see UAD Lecturer Bennett's, Java Wormhole exercise.

4. (Mandatory) From Bennett's Live Neural Network Programming session format,. we will construct a basic artificial neural network from scratch together, without the usage of any machine learning library. (You can follow this guideline of mine here or here, taken from my presentation/programming tutorial at ITS 2019.)

Grab a copy of bluej or use python to follow along to replicate my basic neural network code as I write it out!

5. Upload this code to your github account. Name it "My-Code_For-MLJI-Week-1", or something similar.

• Even after you have done this entire course, practice this basic neural network programming session every 6 to 8 months on your own (as I do here). More practice will grant you more intuition and grasp of Neural Networks/Machine Learning overall.
• As discussed prior, neural networks are universal problem solvers, and therein, learning to construct them is an optimal way to equip one's mind with core ML ingredients.

1. Try to grind through this 3 hour course on Tensorflow (Sections: "Beginner quickstart, Keras basics,Keras.Load data") . Tensorflow is a popular machine learning library. You can start with this, or resort to other libraries like MXNET.

• You can use any one of these libraries for the subsequent weeks. Note that Tensorflow has the support of Google, so it may be wise to choose this library. MXNET also has the support of Microsoft, so that is not a bad option either.

1. Watch Convolution Neural Networks - explained! by Code Emporium.
2. Implement a Convolutional Neural Network (using any ML library like Tensorflow; See example), based on video above, or otherwise, to solve some problem. Include a document including your name, and 1 page describing the problem you solved.
3. Upload this code to your github account. Name it "My-Code_For-MLJI-Week-3)", or something similar.

Example cases by Universal Ai Diploma Lead Lecturer/Instructor:

1. World's 1st open source covid19 diagnosis tool
2. NVIDIa featured pothole detector
3. World's first open source masked face recognition artificial intelligence app (see also my stored face variant)

1. Watch Recurrent Neural Networks - explained! by Code Emporium.
2. Implement a Transformer, or Recurrent Neural Neural Network (using any ML library like Tensorflow; See example), based on video above, or otherwise, to solve some problem. Include a document including your name, and 1 page describing the problem you solved.
3. Upload this code to your github account. Name it "My-Code_For-MLJI-Week-4", or something similar.

1. Alternative: Create an assistant using free IBM Watson Cloud. This technology uses recurrent neural networks. Sign up for IBM now for free. See guideline by UAD Lecturer G. Bennett.

Example from lead Universal AI Diploma Instructor/Lecturer: https://youtu.be/pxgLJSd3_-s Note: Your experiment does not need to include any 3d agent. Only a browser text messenger is fine.

2. Alternative 2: Create an assistant using free RASA Open Source RNN based chatbot

Example with guideline from lead Universal AI Diploma Instructor/Lecturer: https://godquestbennett.medium.com/universal-ai-diploma-ibm-cloud-assistant-recurrent-neural-network-alternative-e42ce4d02a6d Note: The item above unlike Alternative 1, does not require the use of visa card. (requires only your github profile which you would have had initially)

1. Watch Generative Adversarial Networks - FUTURISTIC & FUN AI! by Code Emporium.
2. Watch "Make a Face Generative Adversarial Network in 15 MINUTES!" by Discover Artificial Intelligence.
3. Implement a Generative Adversarial Neural Network (using any ML library like Tensorflow; See example), based on video above, or otherwise, to solve some problem. Include a document including your name, and 1 page describing the problem you solved.
4. Upload this code to your github account. Name it "My-Code_For-MLJI-Week-5", or something similar.

Alternative:

Instructions: https://medium.com/@godquestbennett/universal-ai-diploma-week-5-generative-adversarial-neural-network-instructions-cf1b554ef92

1. Use any colab gan style transfer code to make a custom image (transfer style to any image of your choice).
2. Sign up for any free NFT space like Mintable, and "mint" aka upload your gan image "art" with a nice title.

Example cases by Lead Universal Ai Diploma Instructor/Lecturer:

1. World's first smart NFT being you can talk to
2. Worlds first open source ai car interior designer

1. Watch this fun introduction to Deep Reinforcement Learning by Arxiv Insights.

2. Remember the self driving car that was launched by Google in Arizona? It uses reinforcement learning.

• Simply try to run this self reinforcement simulation here, based on Tensorflow to get a reasonable idea of how Google Waymo was built.

3. Optionally, read my 6 minute introduction to deep q learning. Deep Q learning is a variant of reinforcement learning.

4. Upload this code with screenshots of your experiments, to your github account. Name it "My-Code_For-MLJI-Week-6", or something similar.

Example case by Lead Universal Ai Diploma Instructor/Lecturer:

1. Artificial Neural Networks + Reinforcement Learning for autonomous cars in Open World 3rd World 'Extortionist' Game by God Bennett

Note: Please try to complete the 1st 5 chapters of the Deep Learning Book by Yoshua Bengio et al. (Bengio is one of the winners of the winners of the 2018 Turing Award, a Nobel-Prize like award for Artificial Intelligence. See the CBS News Report on the aforesaid 2018 Turing Award).

You can practice as I do, by writing out 8+ pages of equations from memory, i.e. start by memorizing/understanding equations from Linear Algebra, then chapter 3, then chapter 4, then each time you write out more and more pages from memory. Alternatively, you could practice by writing out and understanding an equation every single day. You likely need to repeat some equations for better grasp.

This semester concludes with a core review of mathematical notations ranging from Linear Algebra, to Numerical Computation. No code shall be uploaded for this last review.

What do you do when you want to hire a driver for a new driving task?

1. Train a non-driver from scratch, i.e. get Licence, Train for 4 years on road.
2. Hire a seasoned person who had already driven with few errors for 4 years.

• I.e. We can say the driver is pre-trained for driving, and time to train this driver for a driving task will be far less.

In a similar way, to solve a problem amenable to Ai use, it is optimal to find a pre-trained Ai model, typically already trained on expensive Ai gpus for several hours (that can cost up to 3000 usd per unit or per month), like rtx 3090s up to V100s.

In the case of Covid19 diagnosis for eg, since Covid19 is identified as a form of pneumonia, instead of training from scratch, it is optimal to:

1. Find a pre-trained pneumonia model. (One with high accuracy/sensitivity/specificity. Quickly explore several pretrained models and select best.)
2. Integrate model best AUC into your pipeline. (Like a user interface built on top of your Ai model to be user friendly)
3. The above is a similar approach taken by large companies like Google, who employ the usage of Auto-ML to help select best models.

• In your case, you manually explore the landscape of available pretrained models by literally trying each of them out.
• Key is to identify ones that clearly specify their accuracy/senstivity/specificity for many cases. (Eg of resource for pre-trained models found on Github and Kaggle)

1. Search for a problem to solve preferably in Jamaica or otherwise worldwide:

   a. If the problem has not yet been solved worldwide or in Jamaica, you can manipulate any existing python/tensorflow code etc, to solve the problem.

   b. If the problem has already been solved, try to apply whatever you learnt to either achieve the same efficiency, or better, by using your own code assemblies.

2. Note that what ever the result, this code is yours, given that the code used is freely usable. The last semester is optional, but will:

   a. Be a good way to showcase your experience, beyond the scope of the guidelines from prior semesters. (aka show creativity and added experience)

   b. Be a huge plus on your github profile, that employers may appreciate!

3. For this final semester in particular, you may or may not chose to upload your code to github (while naming it as something like "My-Code_For-MLJI-FinalSemester"):

   a. If you plan to form a company around that code/solution, then you may opt to keep your final semester code on a private platform like Google drive instead of Github. After that, you may then add a reference to your portfolio page as an entry/item together with works/items from weeks 1 to 6. (See portfolio section below). If you kept that final-semester code to yourself, it would appeal to employers if you describe it briefly as an entry item on your portfolio, regarding how your new company/solution leverages machine learning to solve the unsolved or insufficiently solved problem you discovered.

   b. You could still choose to upload code to Github, prior to which you would have copyrighted it. In your portfolio, MLJI will tell the public that it is copyrighted in this case. (See portfolio section below.)

   c. Otherwise, you could make your final semester code free to the public, and simply upload it to Github, naming it as suggested in the beggining of item 3.

We'll provide you with your own portfolio page (aka your portfolio URL), at mlj-institute.appspot.com/students/< your-username >/< your-mlji-id >. (See sample portfolio here.)

The portfolio will list all your uploaded works, including your final-semester code (if you chose to upload the final-semester-code). You'll provide descriptions and github links of each work entry uploaded to github on the portfolio.

This portfolio will be accessible from your URL above, and will appeal to your employers!

You may return to the main page from here.