AI for everyone: Understanding Machine Learning
Namaste 🙏,
I won’t be proclaiming, “Hey, guys, the world is changing! We might have R2-D2 or WALL-E robots roaming around in a few days.” However, the truth is, we might not be that far away. Thanks to AI, this futuristic possibility is becoming increasingly believable, thanks to the implementation of neural networks and other advanced technologies.
By the end of this article, you will gain a solid understanding of AI, and you can also implement these technologies in your use cases.
What is Machine Learning?
Computer programs that uses algorithms to analyze data and make intelligent predictions based on the data without being explicitly programmed.
Slightly more formal definition is
- “Machine Learning is the field of study that gives computers the ability to learn without being explicitly programmed.” — by Arthur Samuel, 1959
- “A computer program is said to learn from experience E with respect to some task T and some performance measure P, if it performance on T, as measured by P, improves with experience E.” — by Tom Mitchell 1997
Before we dive into the Machine Learning(ML), it’s essential to grasp the big picture of Artificial Intelligence(AI) because ML is a subset of the AI.
I hope you are familiar with the Venn diagram. If not, let’s cover it in a few seconds.
Explanation of Venn Diagram:
It’s a visual representation using overlapping circles to demonstrate the connections between different groups or sets. Each circle represents a group or set, and where they overlap, it indicates common elements shared between them. It’s a helpful method to observe similarities and differences among various groups or sets.
Let’s see the Venn diagram of Artificial intelligence
As you see in the Venn diagram, ML is a subset of Artificial AI.
ML ∩ AI
While all Machine Learning falls under the vast umbrella of Artificial Intelligence, it’s essential to note that not every instance of AI involves Machine Learning.
Assume we are building a Humanoid robot. In this case, IoT is applied to construct the walking human-like robot, with Machine Learning(ML) acting as the mind or engine that powers the intelligence by allowing systems to learn from data.
In the grand landscape of AI, Deep Learning(DL) is another crucial component. Think of DL as a specialised form of ML, where artificial neural networks, inspired by the human brain, enable machines to perform tasks such as image recognition, Natural Language Processing(NLP), and decision-making at a level that was once thought to be reserved for human cognition.
Types of Machine Learning:-
- Supervised learning
- Unsupervised learning
- Semi Supervised learning
- Reinforcement learning
Let’s understand one by one
Supervised learning
In this, we feed data to the algorithms, In which the data are labeled and we know what our output should like having the relationship between input values “X” and output values “Y”.
Examples:
- Prediction of House Prices
- Email Spam Detection
Unsupervised learning
In this, we feed data to the algorithms, which are not labeled or we can say that we don’t know what our output should look like and there is not any kind of relationship between input variable And output variable. We have to recognise patterns based on the data, for doing so, we have to use different algorithms.
Examples:
- Recommendation Engines (like Google, Netflix)
- Computer Vision (like Object Recognition)
- Anomaly Detection
Semi-supervised learning
In this, the algorithm is trained on a dataset that contains both labeled and unlabelled data. While a portion of the data is annotated with input-output pairs (“X” and “Y”), the rest lacks explicit labels. This approach leverages a combination of labeled and unlabelled data to improve model performance.
Examples:
- Image Classification with Limited Labelling
- Text Document categorization with Labels
Reinforcement learning
Reinforcement Learning (RL) is an area of machine learning that differs from supervised and unsupervised learning in that we don’t provide any data to the model. In the absence of a training dataset, it is bound to learn from its experience. RL is an autonomous, self-teaching system that essentially learns by trial and error. It performs actions with the aim of maximizing rewards, or in other words, it learns by doing to achieve the best outcomes.
Examples:
- Natural language processing
- Robotics
At the end, the goal of AI is to make machines that have intelligence like human, right? I believe that Reinforcement Learning with Deep Neural Networks is a step towards achieving this.
OpenAI has built a bionic arm that can solve Rubik’s cube automatically. Their approach is to train their model in the virtual simulator because if we try to try it on physically initially, it may fail. That means we unintentionally damage the physical components, resulting in potential costs of money and resources. Therefore, it’s better to train the model in a virtual environment. For that, we can use pygame library to create an online simulator. Once it performs better than expected, we can deploy it on the physical apparatus.
By now, I hope you guys all have a theoretical understanding of how machine learning works.
