MLJI Machine Learning Jamaica Institute
This platform has been adapted into the globally accredited Diploma, "Universal Artificial Inteligence Diploma" in partnership with Advanced Solutions Technical institute.
This paradigm occurs wrt RobotizeJA's objectives.
MLJI Curriculum Main Page
- Start as a typical computer user with or without programming skills, and end as an Artificial Intelligence product creator and profit from Ai's predicted 15 Trillion dollar 2030 market cap!
- With 800 million jobs to be automated away by 2030, and Artificial Intelligence market cap at 15 trillion around the same time, become a builder of Ai solutions, and capture your chunk of the exponentially growing Ai market.
- As an MLJI trained person [who happens to be a farmer], building out an Ai farming application to classify your farm yield, is one of thousands of exciting practical/profitable AI apps, through MLJI roadmap. (See relevant Ai app by cuccumber farmer).
- As an MLJI trained person [who happens to be a software dev at a bank], building out an Ai based masked face recognition app, is another one of thousands of exciting practical/profitable AI apps, through MLJI roadmap. (See relevant Ai app by creator of MLJI)
- As an MLJI trained person [who happens to be a nurse or doctor...], building out an Ai based covid19 diagnosis, is another one of thousands of exciting practical/profitable AI apps, through MLJI roadmap. (See relevant Ai app by creator of MLJI)
- ...
Bonus-beyond: Humanity's last invention - Introduction to Artificial General Intelligence (Example: Prof Ben Gortzel)
See course content (15 minutes read): https://github.com/JordanMicahBennett/Machine-Learning-Jamaica-Institute_Curriculum_MainPage/blob/master/data/Introduction%20to%20Artificial%20General%20Intelligence.pdf
- The world's smartest people are switching careers into Ai research and application/productization!.
- According to PwC Global, Ai will yield a 15 trillion dollar market cap by the year 2030. This is around the same time 800 million jobs shall be automated away, according to Bank of America.
- It is thus key to intimately benefit in this automation effort, rather than fail to do so.
The AGE of Artificial Intelligence Season 1, Hosted by Robert Downey Jr
- Brain inspired computer code or smart apps, called AGI or artificial general intelligence (predicted to happen by as soon as 2029 or sooner), will perhaps one day be mankind's last invention! For now though, AGI's predecessor, called artificial narrow intelligence, also called artificial intelligence, can do amazing stuff like diagnose diseases better than human doctors, or give game characters the ability to learn to navigate game environments without human aid!
-
Crucially, the Jamaican government has recognized that artificial narrow intelligence, a part of machine learning, is a threat to jobs. Notably, here's a list of jobs where robots or smart software are already replacing humans.
-
As the 1st machine learning school established in Jamaica/Carribean in 2016/2020, MLJI seeks to help more and more Jamaicans to find out what steps they can take to acquire skills needed to stay relevant in a world that's more and more driven by artificial intelligence, that shall replace more and more jobs with physical robots and or smart software.
-
As such, MLJI is an institute that aims to bring more Jamaicans into the modern world of artificial intelligence and programming through fun, tailored lectures. Pertinently, see the "why" tab on the MLJI website, to find out why high-schoolers, or even busy Masters/PhD/Professor people who are not trained in the machine learning field can participate in modern machine learning!
This curriculum is designed to cater to two main categories of users, including non-programmers and programmers. Those two categories naturally cover persons including those with High school diplomas, Bachelors degrees, Masters degrees, PhDs, and Professorships.
The two main categories above are expressible as 2 types, based on the goals that MLJI desginates its students to perform.
As such, from here on out, you select one of the types below, and begin your learning process:
By the time you complete either one of MLJI's curriculum types above, you would be a marketable candidate, qualified to attain remote work with the capability to earn beyond 80,000 USD per year, whether you're a high schooler or professor who was not priorly trained in Deep Learning. Also, you would be in far better position for local work, and better able to prepare for automation of more and more jobs in the globe, and in the country.
- Candidates must have at least passed the GSAT or completed the PEP or something similar. (From Primary to Tertiary Education and beyond, i.e. Additionally, up to PhD/Professor level candidates are welcome too!)
- The requirement above essentially concerns just before high school level all the way up to PhD/Professor people who want to profit from Ai development!
- Candidates must be willing to devote at most 6 to 8+ weeks to MLJI's curriculum. In other words, candidates must be willing to build himself/herself, by acquiring machine learning skills through MLJI's 6-8+ week curriculum.
- Candidates need to have access to machine learning computers, or online machine learning compute like free colab instances.
- This course also advises for the use of datasets such as those from kaggle. Ai/Machine learning is normally complete with ingredients:
- Ai Knowledge/Skills (which this course builds in each candidate)
- Ai Compute
- Ai Compatible/Consumable Datasets
As underlined in prior Live Neural network sessions by organizer Bennett, artificial neural networks can represent a wide variety of utilities/functions as seen in Universal Approximation Theorem, and since neural networks power most cognitive/smart apps today, ranging from Tesla's self driving cars, to Facebook's auto face tagging Ai, MLJI targets this core field. Learning neural networks which are general models, gives rise to intuition needed to venture elsewhere if neccessary.
Why combine artificial neural networks from scratch in with focus in practical machine learning library stack such as tensorflow/pytorch...?
⚫Pertaining to artificial neural networks from scratch in particular, as Microsoft's Joseph Albahari says (in his csharp based neural networks from scratch lecture), machine learning libraries are reasonably a great place to end up, but not necessarily a great place to start!
As noted by Bennett, understanding these broad/fundamental neural network models from scratch without machine learning libraries, builds intuition in applying and debugging machine learning libraries.
📝In addition to the above, this reasonably smart curriculum includes "dynamic pathways" assuming some or no programming knowledge in addition to artificial neural networks from scratch. (The insight (that application of neural networks from scratch is a fundamental tool), of others like Sentdex and Joseph's, requires prior programming knowledge, and pertains to learning Neural Networks' internal structure, while this curriculum accommodates candidates with no programming knowledge etc)
📝This Diploma combines artificial neural networks from scratch, in a beyond Joseph's code based csharp lecture, in an illustrative class diagram based Java paradigm designed for new candidates, (around 1000 lines of code without machine learning libraries), with primarily practical detailed use of python aligned machine learning library enabled projects/via tensorflow for eg, therein helping candidates to gain intuition in the usage of machine learning libraries like tensorflow. University programs tend to confluence both Java and Python, where this Diploma focusses primarily on python.
📝As usual, as underlined in prior Live Neural network sessions by organizer Bennett, artificial neural networks can represent a wide variety of utilities/functions as seen in Universal Approximation Theorem, and since neural networks power most cognitive/smart apps today, ranging from Tesla's self driving cars, to Facebook's auto face tagging Ai, MLJI targets this core field. Learning neural networks which are general models, gives rise to intuition needed to venture elsewhere if necessary.
This course has been adapted for AICE via AdalabsAfrica, and Delaware State University.
God Bennett is the founder of MLJI, author of "Artificial Neural Networks for Kids" (See either Amazon url or free copy) and inventor of the "Supersymmetric Artificial Neural Network".
- Key Fundamental Artificial Neural Network Coding (Without bases like this that explore Universal Function Approximators like Neural Networks, it is harder to debug Machine Learning Libraries when something goes wrong!)
- Practical Convolutional Artificial Neural Network application (Seen in real world in Self driving cars, Disease Diagnosis engines etc)
- Practical Recurrent Artificial Neural Network application (Seen in real world in any voice recognition App)
- Practical Generative Adversarial Network application (Seen in real world in synthetic Avatar Apps)
- Practical Reinforcement Learning application (Seen in real world in Electricity Consumption optimization)
- Practical Data Wrangling (Without Data Ai is essentially meaningless)
- Deep Learning/Paper Writing (Exposed in Certificate)
- Ai Product Building (Exposed in Diploma)
- Rapid Ai Product Building Methodology (Exposed in Diploma) (including schedules)
- Artificial General Intelligence Basis (Exposed in both Certificate & Diploma. )
- Separate/outside of the Syllabus-Exam rubric & Diploma, UAD lecturer Bennett uniquely introduces Human Level Ai/Artificial General Intelligence - (predicted by around 2023-2029 or later, by experts to be mankind's last invention).
See a reasonably soft Introduction to Artificial General Intelligence
MLJI's course is designed for teens/adults. Albeit, take a look at these intriguing books for babies: Einsteinian books for babies
The most rigorous attempt to create an explanatory framework for Deep Learning (about 500 pages), recently built heavily in Physics: https://deeplearningtheory.com/PDLT.pdf
See also - Short summary: "NOW IT’S OFFICIAL: WE DIDN’T UNDERSTAND HOW NEURAL NETWORKS WORK TILL NOW".