Machine Learning: The Inception of Machine Intelligence
The ability to learn from experience is what makes us humans. When we were little children, we almost knew nothing. As we got older, we were able to tell the difference between an apple and a banana. But how?
Through learning and experience, we can recognize that a banana is yellow and long, and an apple is red and round. The same concept applies to machines.
With machine learning, we aim to help machines identify, what an apple is and what a banana is. To achieve this, we have to train the machines.
Okay, first we will list out all the features of banana and apple, such as their color, shape, and hardness, the features can be anything as long as it helps to identify the fruits independently. Keep in mind that as the number of features increases, the accuracy of the machine also increases.
After inputting these data, we will train the machine to identify each fruit separately. At first, it might make some mistakes, but as we train it, the machines will improve on their own and learn from their experience to tell the difference between an apple and a banana. This is called machine learning.
Artificial intelligence Vs. Machine learning
Artificial intelligence is a topic that is well discussed in the past few decades. Real-life bots like Sofia and Asimov have made so many people curious and intrigued by artificial intelligence. Trust me when I say that I am one of them. I love how cute these robots look, except for Sofia though. She creeps me out.
Coming back to the point, There is a common misconception that artificial intelligence is the same as Machine learning, But it’s not the same.
Artificial intelligence is a vast field that contains so many other subfields like, Natural language processing (NLP), Robotics, Deep learning, Computer vision, and a lot more. Machine learning is just a subsequent field of artificial intelligence that helps a machine learn from experience. Machine Learning enables ‘machines’ to make decisions based on their previous experiences.
Categories in Machine Learning
Humans can make decisions based on their previous experiences, and that’s one of the factors that make us intelligent. Using machine learning, we teach a robot how to make decisions on their own from previous experiences, that too without any human intervention or instructions.
Machine learning is a field of artificial intelligence, that allows a machine to learn from its experience rather than being explicitly programmed. The field is interested in the development of computer algorithms for transforming data into intelligent actions and is called machine learning.
Based on how the machine learns, machine learning is classified into four types.
- Supervised Learning
In supervised learning, as the name verses, machines are trained with well-labeled data under supervision. We supervise or guide the machines with labeled data that already contains some correct answers. After the training, the machines are given, a set of training samples to analyze the supervised training algorithm.
For example, Imagine you are given a basket full of fruits. The first step is to train your machine. To do that, you need data.
The machine is taught that the apples are red, round, and hard in texture. The bananas are long, yellow, and soft in texture. The same process is carried out with all fruits. With this data, the machine is trained. The next step is to test the machines. Since the data was already labeled, the machine can easily identify the fruits and put them in each of their baskets. The machine learned from the data and applied that knowledge to categorizing the fruits. This is called supervised learning.
Although supervised learning has a high accuracy level, the data has to be hand-labeled by a data scientist or machine learning engineer, so it can end up being costly.
- Unsupervised Learning
In unsupervised learning, the machines are taught non-classified and unlabeled data. This kind of method helps machines learn on their own without supervision. Here, the task is to group unlabeled data based on their patterns, similarities, and differences.
For example, a set of pictures of dogs and cats are presented. The machine has to identify which is a cat and which is a dog. Based on that information, the machine groups the pictures into cats and dogs. Since there is no previous training involved, the machine has to identify which one is which, on its own.
The problem with unsupervised learning is that the accuracy of the acquired result won’t be high and the application spectrum is limited.
- Semi-supervised Learning
Unsupervised algorithms and supervised algorithms have their demerits. To tackle these, the semi-supervised algorithm was introduced.
A semi-supervised algorithm uses both supervised and unsupervised algorithms to train the machine. Using all labeled data can be expensive and, using all unlabeled data can end up in low accuracy. But a mix of both can do the job. So in semi-supervised learning, we will use a large amount of unlabeled data and a small amount of labeled data. The basic idea behind this algorithm is that first, the programmer will cluster similar data using unsupervised learning algorithms and then use the labeled data to label the rest of the unlabeled data.
The concept behind all three of these algorithms is similar to a student and a teacher. A student is taught concepts under the supervision of the teacher, both at school and at home. This is known as supervised learning.
In unsupervised learning, the student is supposed to learn the concept by himself. In semi-supervised learning, the teacher inculcates, a concept in the student, and homework is assigned based on the same concept.
- Reinforcement Learning
In supervised learning, we will use an answer key to train the machine. In reinforcement learning, we don’t use any answer key. Rather, the machine learns on its own from its previous experience, and on the successful completion of the task, the machine, is rewarded.
This is just like how we train our dogs. When we command the dog to sit, he is rewarded with a treat. So the next time, the dog knows he will get a treat if he completes the task. So he will complete the task to get the treat.
Applications of Machine Learning
- Traffic Prediction
Whenever we want to visit a new place, we depend on our one and only trustee google map. Have you ever noticed how the google map predicts traffic?
Well, that’s an application of machine learning. It helps to detect if the traffic has cleared, is slow-moving, or is heavily congested. This is done based on the real-time location of the vehicle and the average time taken.
- Image Recognition
Have you ever wondered how you got that auto-tagging suggestion from Facebook?
Well, that’s another magic of machine learning. Image recognition is one of the frequently used applications of machine learning and is used to recognize an object, a person, a place, a digital image, etc. Facebook uses machine learning for image recognition and face detection algorithms for tagging suggestions.
- Speech Recognition
If you are lazy like I am, you must have found it easy to use virtual assistants, like Google and Siri. These types of virtual personal assistants use speech recognition to convert voice instruction to text commands. And that’s another application of machine learning.
- Self Driving Cars
The Google self-driving cars have always been a topic that made people curious. But how does the car work? Machine learning uses a significant role in self-driving cars. The companies like tesla are working on their tesla autopilot car that uses unsupervised ‘learning’ to detect humans, cars, and objects while driving.
- Medical Diagnosis
Medicine is the other area where machine learning has shown its prominence. Medical diagnosis is an application of machine learning and predicts if a patient has any chronic illness based on their medical history and reports. Machine learning can also be used in predicting brain tumors and brain-related diseases effectively.
Machine learning is a field that’s continuously growing and, there is always more to know. If you are a techie and have a curious mind to learn more. Then stay tuned and, don’t forget to check out our other blogs.
Until next time, Adios…