Free Coursera Offerings - A Review - Centre Of Robotics Excellence

COVID and Online Education

As Schools all across the world scrambled to go online with lockdown measures imposed across various countries, the state of readiness of each institution became clear. However, unlike many industries where digital transition was with little successful precedent, education has a relatively good track record with Youtube channels spanning every conceivable discipline. More recently, we saw the rise of for profit MOOC or Massive Open Online Course providers such as Udacity and edX. Running on a dedicated platform, providers can not only offer passive video content but too, increase interactivity through the use of student forums, quizzes and other web-based tools.

It comes as no surprise that COVID-19 has brought about a unique opportunity for growth for MOOC providers. Not only are students and educators forced to embrace online learning and make it work, a sizeable portion of the workforce has been forced to question the adequacy of their skillsets as pandemics and periods of lockdown may become “the new normal”. To that end, MOOC providers have been offering various goodies to entice users to experience their platforms. While their intentions may truly be altruistic, there is undoubtedly something to be gained through unprecedented exposure levels and the lock-in of longer term offerings.

Coursera’s Offerings

Coursera’s free online offerings caught my eye as they extended till the end of July, whereas other providers such as Udacity were offering up to a free month of courses. In particular, I was interested in the courses promising to teach cloud tech skills. While I’ve had some experience with IoT Dashboards such as Freeboard.io, Google’s Firebase and Web Scraping for Machine Learning data with Microsoft Azure, I’ve never had a chance to learn these skills formally and could not pass up the chance to see how others are developing curriculum around this much-hyped technology.

Due to time constraints, as well as the inanity of going through every single course offered (a lot of them covered the same topics with different platforms), I decided on 3 which would allow me to gain an insight into the 3 brand-name platforms with sufficient grounds for comparison.

Additionally, I also tried my hand at Computer Vision Basics offered by the University at Buffalo and The State University of New York as I run an Introduction to Computer Vision with Python course myself. However, I will not be going through that in this review as it covers a very different topic. Suffice to say, it really is the basics, and is based on Matlab, which is used more for academic purposes than real world deployments. That said, a complete beginner may find explanations of concepts useful.

Machine Learning for Business Professionals (Google Cloud)

Course Description

This course is intended to be an introduction to machine learning for non-technical business professionals. There is a lot of hype around machine learning and many people are concerned that in order to use machine learning in business, you need to have a technical background. For reasons that are covered in this course, that’s not the case. In actuality, your knowledge of your business is far more important than your ability to build an ML model from scratch. By the end of this course, you will have learned how to:

Formulate machine learning solutions to real-world problems
Identify whether the data you have is sufficient for ML
Carry a project through various ML phases including training, evaluation, and deployment
Perform AI responsibly and avoid reinforcing existing bias
Discover ML use cases
Be successful at ML

Review

I have no complaints about this course! From the get go, the description is honest and welcoming. This is not a course that aims to teach you the technicalities of Machine Learning. Rather, it attempts to empower the user by showing you how easy it is to implement your own applications which utilize common Machine Learning Algorithms in various contexts. Unfortunately, there is a limit to this approach and students should not expect to be able to formulate or implement complex pipelines, which may be required in some applications. However, as a beginner, the course and the platform does provide simple rules of thumb (minimum of 10 data points per category for classification, 80/20 training/test splits etc.) while guiding students through basic projects end-to-end on Google Cloud. By the end of the course, students will be able to build applications, even if they may not be perfect. This is valuable and breaks down an important mental barrier. Even more impressively, there is a significant amount of time spent on the limitations and ethics of Machine Learning. This stems from an understanding that technology is a powerful tool, and knowing the benefits and not the costs may result in undesired consequences- an issue that I wrote about on Medium.

Getting Started with AWS Machine Learning

Course Description

Machine learning (ML) is one of the fastest growing areas in technology and a highly sought after skillset in today’s job market. The World Economic Forum states the growth of artificial intelligence (AI) could create 58 million net new jobs in the next few years, yet it’s estimated that currently there are 300,000 AI engineers worldwide, but millions are needed. This means there is a unique and immediate opportunity for you to get started with learning the essential ML concepts that are used to build AI applications – no matter what your skill levels are. Learning the foundations of ML now, will help you keep pace with this growth, expand your skills and even help advance your career.

This course will teach you how to get started with AWS Machine Learning. Key topics include: Machine Learning on AWS, Computer Vision on AWS, and Natural Language Processing (NLP) on AWS. Each topic consists of several modules deep-diving into variety of ML concepts, AWS services as well as insights from experts to put the concepts into practice.

Review

With an estimated completion time of 8 hours (as opposed to 12 hours for Google Cloud’s offering), I found the course description to be disingenuous and formulaic, rattling off statistics pertaining to AI (of which ML is a subset). Additionally, MOOCs have their limitations (at the moment), which I’ve footnoted. However, the course did start off well, developing knowledge and vocabulary for the first 2 weeks (great for beginners, making clear the distinction between AI/ML/DL/NN) and introducing students to the Machine Learning Pipeline. The second half of the course focused on AWS products, introducing Rekognition, DeepLens, Comprehend, Translate, Transcribe, Sagemaker, GroundTruth, Neo, and Glue. Yup, the class was mired in AWS product-specific names, and connecting nomenclature with functionality might be difficult for some, especially when different platforms adopt seemingly arbitrary demarcations between products. While individual lessons were pretty clear, the lack of mandated examples (and along with it, the ease of access to the platform), makes for an unengaging experience, much like watching a shopping mall cooking demonstration looking to sell the latest non-stick pan, and expecting one to begin their journey as a chef. Unfortunately, the course as a whole felt disjointed, and felt like a compilation of topical talks rather than a purpose built curriculum. This creates cognitive load in understanding flow. However, the ease of use of the platform does make any shortcomings in course delivery slightly more forgiving.

Developing AI Applications on Azure

Course Description

This course introduces the concepts of Artificial Intelligence and Machine learning. We’ll discuss machine learning types and tasks, and machine learning algorithms. You’ll explore Python as a popular programming language for machine learning solutions, including using some scientific ecosystem packages which will help you implement machine learning.

Next, this course introduces the machine learning tools available in Microsoft Azure. We’ll review standardized approaches to data analytics and you’ll receive specific guidance on Microsoft’s Team Data Science Approach. As you go through the course, we’ll introduce you to Microsoft’s pre-trained and managed machine learning offered as REST API’s in their suite of cognitive services. We’ll implement solutions using the computer vision API and the facial recognition API, and we’ll do sentiment analysis by calling the natural language service.

Using the Azure Machine Learning Service you’ll create and use an Azure Machine Learning Workspace. Then you’ll train your own model, and you’ll deploy and test your model in the cloud. Throughout the course you will perform hands-on exercises to practice your new AI skills. By the end of this course, you will be able to create, implement and deploy machine learning models.

Review

Week 1 content started out relaxed, and made good use of Jupyter Notebooks on Azure to get users started on the basics of Python. I was expecting a lot from this course as a result, and was hoping that it would go slightly more in-depth and provide more opportunities for coding than Google’s offering. However, the entirety of Week 2 was dedicated to organizing ML teams with very formal frameworks, making me question the target audience for this course. Surely the designers were not expecting students who needed introduction to the most basic Python concepts such as math and logical statements to lead large AI teams? The later weeks got students back on track by going through Azure’s offerings, yet I couldn’t shake off the nagging feeling that this course was not well thought out. The course was generally less well integrated than the Google offering, with whole video walkthroughs presented before leaving students with a Notebook to mess around with (little direction is given). That might have been alright if the walkthrough was done well. Unfortunately, the course instructor reads everything, word by word, from menu items on websites, to import statements in Python, to RAM configurations of virtual machines. Additionally, I am of the opinion that watching someone write code is not a replacement for explaining the pipeline and its components clearly, which was what is done most of the time. Even at 16 hours, the longest amongst the 3 courses, it is naïve to think that a student will be equipped with all the necessary skills having known nothing at the start. To achieve even mastery of Python within this timeframe would be a miracle. A false sense of knowledge is a dangerous thing.

Conclusion

It should be clear that between the 3 courses reviewed, the one that stood out for me is Machine Learning for Business Professionals by Google Cloud. Yet this isn’t for everyone. Those who are looking to build novel applications with, or do research on machine learning, should look to start off on other resources such as the deeply popular Machine Learning course taught by Andrew Ng (the co-founder of Coursera) which is free to audit. After all, Google’s offering is targeted at Business Professionals and does a great job at that.

Would I say that there is no value in going through the other 2 courses? Not quite. For those who already have a background in one platform and are looking to transition to another, these courses may act as tutorials to understand the platform (especially their idiosyncratic naming conventions), with step-by-step guides on how to go about achieving specific goals. However, I am still of the opinion that both leave much to be desired in terms of structure and content, and attempt to do too much with too little time and resources. One might be better off watching specific videos in the course, or through another platform.

At the end of the day, these courses are provided at no cost to anyone by Coursera. Sure, it might not be perfect but for many, it will represent the first steps to be taken in a world growing in uncertainty. For that, we have to thank Coursera, as well as every other provider of online education, for lowering the barriers to entry in acquiring a new skill. If there were to be a silver lining to COVID, I would hope for it to be the pedagogical advances in online education brought about by its forced, rapid uptake en masse.

Footnote

For the most part, MOOCs are wholly insufficient for someone to begin work in an industry. However, they do serve as good tasters to gain a basic understanding of concepts to start figuring out how to advance their skills.

Blog