Introduction to Machine Learning

10 years ago, Rober Downey Jr. playing Iron Man is what got me interested in Artificial Intelligence (AI). He is my inspiration for Machine Learning and AI. But first, What is AI? And how do Machine Learning (ML), Deep Learning (DP), and Natural Language Processing (NLP) come into AI?

ML, DP, and NLP are subsets of AI; Machine Learning is a subset of AI, Deep Learning is a subset of Machine Learning, and Natural Language Processing overlaps ML and DL. Natural Learning Processing uses all these techniques, among other things. 

What is Machine Learning?

“Machine Learning is the field of study that gives computers the ability to learn without being explicitly programmed” Arthur Samuel, 1959.

The definition of Machine Learning hasn’t changed since 1959, but what has changed is the computing power and the way we handle the data.

What’s the difference between Machine Learning and Python? Traditional Programming works in a rule-based model, while in Machine Learning the program will learn itself from a set of data we provide.

The engineering definition of Machine Learning as defined by Tom Mitchell in 1997 is: “A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improves with experience E.” 

Ergo, it is a computer program that learns from the data that we provide. The data is the only information we have that we feed the program. With respect to some task T, that would be the solution we are trying to get. Performance measures P is related to the Machine Learning model itself. When we create a Machine Learning Model and train it with the data, we need to make sure it is trained correctly, hence this is where performance measures come in. It is the metric to analyze our model.

We create a model, we train the model with a set of data, then we use another set of data which helps the model predict the solution we are looking for. We improve the model by checking the performance measures. 

Machine Learning can be seen everywhere. It’s in even your pocket! A lot of applications on your phones use Machine Learning; If you are using g-mail, when you type an email, there is an auto-suggestion to complete your sentence, that is machine learning. Even when you forget to write a subject line, Machine Learning suggests the best subject line for your email.

Amazon Go uses Machine Learning along with other technologies to allow a queue-less grocery store. There is no cashier. You scan a QR code, take what you need, and leave. The different technologies used to calculate the price of the things you picked up and they charge your wallet.

Netflix also uses Machine Learning by suggesting movies and shows based on what you previously watched. Over 75% of what people watch on Netflix comes from recommendations. 

Airbnb uses Machine Learning for a lot of things too. It suggests the appropriate pricing for hosts and helps customers find the right place for them. 

Machine Learning is everywhere. AI is the future. Whether you know it or not, you are already using Machine Learning.

When to use Machine Learning?

  • Problems for which existing solutions require a lot of fine-tuning or a long list of rules: one Machine Learning algorithm can often simplify code and perform better than the traditional approach.
  • Complex problems for which using a traditional approach yields no good solutions: the best Machine Learning techniques can perhaps find a solution.
  • Fluctuating environments: a Machine Learning system can adapt to new data.
  • Getting insight into complex problems and a large amount of data.

Types of Machine Learning

  • Supervised Learning can be used for predictions. When you want to predict something these are the techniques to use. This is applied when you are already giving the input and the output to the model. The input being features and the output being labels.
  • Unsupervised Learning can be used to get insights into the patterns which were unknown before. You add data and the model will find the patterns. It is mainly used by Deep Learning techniques. Here we don’t know the output or labels. There are some sets of problems, and we have the data, but we need to get information out of this data. 
  • Reinforcement Learning can be used for the action. It’s mainly used in the gaming and robotics world.

Types of Supervised Learnings

  • Classified: A classification problem is when the output variable is a category, such as “red” or “blue” or “disease” and “no disease”. The only thing we want our model to tell us here is in which category does the set of features belong.
  • Regression: A regression problem is when the output variable is a real value, such as “dollars” or “weight”.

Types of Unsupervised Learnings

  • Clustering: A clustering problem is where you want to discover the inherent groups in the data, such as grouping customers by purchasing behaviour.
  • Association: An association rule learning problem is where you want to discover rules that describe large portions of your data, such as people that buy X also tend to buy Y.

7 Step Machine Learning Approach

  1. Data Collection: We have to collect data such as information, inputs and outputs depending on the industry and type of problem you want to solve.
  2. Data Preparation: Once the data is collected we have to prepare the data by cleaning it from all the excess noises or unwanted things.
  3. Choose model: We have to pick the model to which we will be giving all the data we have. We don’t have to create a model, we choose the model based on the problem we want to solve.
  4. Train Model: Once we choose the model and have the data, we train the model and add the labels.
  5. Evaluate Model: We need to make sure the model is trained well and we use performance metrics to evaluate the model.
  6. Parameter Tuning: If the model isn’t working well we then tune it.
  7. Predict: At the end, once the model is working well and we add the inputs we have it predict our solution/answer.

Application of ML

  • Analyzing images of products on a production line to automatically classify them
  • Image Classification
  • Convolutional Neural Network (CNN)

To analyze images there are certain models to solve these kinds of problems such as CNN.  

  • Detecting tumours in the brain
  • Semantic Segmentation (CNN)
  • CNN

We can solve these kinds of problems using Semantic Segmentation.

  • Segmenting clients based on their purchase. You can design different marketing strategies for different segments
  • Clustering (K means)
  • Unsupervised

Why should you learn ML?

  1. Machine Learning helps increase your efficiency.
  2. You can understand your customers better.
  3. You can personalize your marketing campaigns
  4. Machine Learning recommends products to your customers
  5. Machine Learning helps to detect fraud.
  6. Learning ML brings in better career opportunities
  7. Machine Learning Engineers earn a pretty penny
  8. Machine Learning Jobs are on the ride.

To learn more in depth about Machine Learning and all its advantages, watch the full video of the mini class below.

Register now for our Data Science Course and learn more.

Related Articles

Responses

Your email address will not be published. Required fields are marked *

Ready to start your Data Science Journey? Join our Data Science & Machine Learning BootcampLEARN MORE