Original article was published on Deep Learning on Medium
Machine Learning, AI, Deep Learning are buzz words being heard daily on TV, in workplaces, at gatherings, etc. Maybe you’re a little bit embarrassed to ask what’s Machine Learning or AI, or maybe you have the wrong understanding of Machine Learning. Either way that’s okay because this article serves as an introduction to Machine Learning, I wrote it in a Q&A format so it becomes easy to follow and understand.
What’s Machine Learning?
Machine Learning is a subset of Artificial Intelligence (AI) and it’s about writing software codes to enables computers (or machines in general) to get better at a given task on their own without human intervention.
Some people argue that Machine Learning is a fancy way to say “Statistical Analysis” which is the science of collecting data and uncovering patterns and trends.
Either way, think about all the data being generated daily and how people try to make sense of it to make their lives better, that’s Machine Learning.
For example, if someone steals your credit card and makes an online transaction. You will probably get an email or text from your bank asking to verify this transaction otherwise the bank will consider it fraud. Your bank’s algorithm (the machine) learned your credit card purchasing habits through your purchase history and when out of the norm transaction was detected the bank suspected it’s a fraud. This is a form of Machine Learning and probably it’s decision tree Classification.
Another example is a car company trying to predict sales for next year based on this year’s numbers and historical data, that’s a form of Machine Learning and could be linear Regression.
Also, based on your creditworthiness some times a bank sends an offer to apply for a low-interest loan or credit card but doesn’t send the same offer to your friend who doesn’t have as good credit. That’s a form of Machine Learning that could be Clustering.
Finally, self-driving cars with auto-pilot roaming around the city on roads it never has seen before, that’s a form of Machine Learning called Reinforcement Learning.
There are many more examples of Machine Learning in our daily lives but we don’t usually notice it.
If Machine Learning is a subset of AI, what’s AI?
Artificial Intelligence is the science of making computers behave like humans in terms of making decisions, text processing, translation, etc. AI is a big umbrella that has Machine Learning and Deep Learning under it. Every Machine Learning algorithm is considered AI but not every AI algorithm is considered Machine Learning.
For example, Google translate is based on AI called Natural Language Processing (NTP). The Amazon Alexa is based on multiple AI services first to understand you, apply sentiment analysis and figure out what you said, search for an answer, and come back with an audio reply.
The facial recognition on your smartphone is also an AI using the Convolution Neural Network (CNN).
Deep Learning is a subset of Machine Learning that tries to mimic the human brain in terms of using neurons and layers to learn. In Machine Learning you have to tell your algorithm what features to learn but in Deep Learning the features will be extracted automatically by the algorithm.
What are the different types of Machine Learning?
Machine Learning can be divided into 3 categories.
Supervised Learning You give the algorithm labeled data and the algorithm has to learn from it and figure out how to solve future similar problems. Think of it as if you’re giving the algorithm problems and answers, the algorithm has to learn how these problems were solved in order to solve future problems in a similar manner. This is like the above example where the bank learns from your habits which credit card transactions are legit and which are fraudulent.
Unsupervised Learning You give the algorithm a problem without any labeled data or any prior knowledge of what the answer could be. Think of it as if you’re giving the algorithm problems without any answers, the algorithm has to find the best answer by driving insights from the data. This is similar to the clustering the bank did to its customers according to various parameters and decided that some are eligible for a credit card offer, others for a line of credit offer, and others aren’t eligible for anything. This is usually done using a Machine Learning method called K-Means.
Reinforcement Learning This is when the algorithm learns from its own experience using reward and punishment. The easiest example is self-driving cars.
Why suddenly Machine Learning became a hot topic?
Machine Learning has been here since the ’50s or maybe earlier. The more recent emphasis on Machine Learning is due to the surge in computer processing power and the drop in chipset prices. Think about the smartphone in your pocket and all the processing power in it, this didn’t exist in the ’90s or early 2000.
All these new powerful affordable chipsets of CPU’s and GPU’s made Machine Learning economical and easily accessible to everyone. With the cheap chipsets, high adoption of the internet, and open sources everyone can use the power of Machine Learning to solve everyday problems.
How big companies use Machine Learning and AI to better serve their customers?
The best example is the Collaborative Filtering used by almost all major companies, which is basically “Since the customer bought this, he would be interested in that?”.
Think of Netflix and how they recommend shows for you to watch or Amazon and how they recommend products for you to buy, usually they get it right. This recommendation is collaborative filtering and there are many algorithms powering it, the Apriori Algorithm is one of the famous ones.
Is Machine Learning complicated enough for individuals and small businesses to take advantage of, and is it true that only big companies are benefiting from it?
This is way far from the truth, in fact, all big tech companies are trying to democratize Machine Learning and make it accessible to everyone whether they have programming knowledge or not.
Take Amazon AWS for example, they offer AWS Transcribe that can convert speech to text, AWS Comprehend that does sentiment analysis, AWS Rekognition that recognizes objects and faces, AWS Polly that can read out something to you. All these services need just a few clicks and no programming knowledge at all. You can be up and running with these services in a few minutes in addition to that most of the Cloud providers offer a free tier that you can use.
Recently AWS released AWS Forecast where a small business can use it to get sales forecasts and other data insights about their business’s sales. The interesting thing about AWS Forecast is that it’s the same exact algorithm used on the Amazon.com website. Just take a second and imagine the power of an algorithm used on Amazon.com to recommend products and predict sales for millions of customers and how they are giving it to small businesses and individuals. The point is that AI & ML are easily accessible to everyone.
Google Cloud and Microsoft Azure offer very similar services, they literally make AI available to everyone. No programming skills needed, no need for you to know what’s happening under the hood, just use it and enjoy it.
What are the available tools to build Machine Learning models?
Machine Learning is about building algorithms so you can do that with any programming language you’re comfortable with. The two most common Machine Learning languages are Python and R. There are many Machine Learning libraries in these two languages, a very famous one is Scikit-learn.
But again you could leverage the cloud provided AI tools or if you have a bit of programming knowledge you can download ready-made models along with instruction from open-sources like GitHub.
How to assess the performance of a Machine Learning model?
That’s a big topic and depending on the Machine Learning type. For example, in Supervised learning Regression and Classification algorithms, we use the Confusion Matrix. This Confusion Matrix along with parameters like Recall and Precision can tell us how the model performed by comparing actual values to predicted values.
What are Type 1 Error and Type 2 Error?
Type 1 error is when your algorithm makes a positive prediction but it’s negative. For example, your algorithm predicted a patient has cancer but in fact, he doesn’t. That’s a type 1 error.
Type 2 error is when your algorithm makes a negative prediction but it’s positive. For example, your algorithm predicted a patient doesn’t have cancer but in fact, he does. That’s a type 2 error.
As a beginner, How do I get started with Machine Learning?
The answer depends on what you want to do with Machine Learning.
If you want to become a data scientist and develop Machine Learning models then you can start by some online classes. Websites like Udemy, Udacity, and Coursera offer an abundance of non-expensive Machine Learning courses for all levels from beginners to experts. Then you could apply your knowledge towards some Kaggle Machine Learning competitions.
If your goal is to understand Machine Learning in general and its effect on your business while getting a certificate from a reputable University, then check online certificate programs at universities like Stanford, Colombia, Wharton, UC Berkely, etc.. Almost every major University offers executive online certificate programs tailored for leaders in the executive and senior management positions.
The point is first to decide what you want to learn about Machine Learning then take that path. Tons of learning available online that can help everyone take their preferred Machine Learning journey.
Is Machine Learning taking my job away?
This is a big debate, the short answer is that Machine Learning is changing lives and will continue to change lives for the better. Big advancement is happening in the field of Machine Learning through open-source collaboration and falling prices of powerful chipsets. These advancements will keep getting better daily.
With every technology leap, people’s lives and jobs changed. The classical example of ATM machines when they started to become popular many were afraid that all tellers will lose their jobs. However, what happened is that tellers moved into better high paying jobs so instead of handing you cash over the counter they became mortgage and investment consultants along with other jobs in the bank. This benefited both the end customer in terms of more available personalized service, fast cash withdrawals and the employees in terms of higher salaries and commissions.
So, some people’s jobs will change because of Machine Learning but if history is any indicator this change will be for the better of everyone. AI and Machine Learning are already making our lives better, just take a look at your smartphone and think how much you can do with it, or maybe think how little you can do without it. Most services in your smartphone are AI-powered.