Machine learning

Machine Learning

What is Machine learning?

Technology is advancing at a very high rate in today's world, we have used it to create things that can help us learn new things and to satisfy our needs. Now, let's take it to the next step by making our machines more sophisticated. Usually we have to tell a machine to perform a task, but what if we could make it execute that task on its own? 

Machine learning is the scientific study of algorithms that a computed uses to perform a specific task without requiring instructions. It relies on patterns and inference to perform. The computer in machine learning uses a mathematical model to analyze data to make predictions without having to be asked to perform such action and it is also a form of artificial Intelligence or AI for short. In other words, its a computer that works independently, like all the robots you see in movies about science fiction. 

What is AI?

In computer sciences, AI or Artificial Intelligence is intelligence demonstrated by a machine or computer and it is compared to human intelligence. It also the study of intelligent agents any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals or completing its tasks. the term artificial intelligence is often used to describe machines that mimic cognitive functions that humans associate with the human mind in things such as learning and problem solving.

We have different types of AI called strong AI and weak AI. Weak AI are the types of AI that work more like an assistant or task manager. Google assistant is an example of weak AI it works to manage your schedule, set up reminders, control the lights on the living room. It is designed to deal with the mundane tasks of our lives. Now, strong AI is something out of science fiction. Strong AI is the type of AI that can perfectly replicate how the human brain is. It is unfortunate that we have not reached these type of technology because to fully perfect it we have to understand how the brain works.

Deep Learning

Deep learning is an artificial intelligence function that imitates the works of the human brain in processing data and creating patterns for use on decision making. Deep learning is a sub-set of machine learning in artificial intelligence that has networks capable of learning unsupervised from data that is unstructured or unlabeled. It is also known as deep neural learning or deep neural network.

Deep learning has evolved hand-in-hand with the digital era and has brought an explosion of data in all forms and from every region of the world. This data, known simply as big data, is drawn from sources like social media, internet search engines, electronic commerce platforms, and online cinemas. This amount of data is accessible and can be shared through applications like cloud computing.

Labeled and Unlabeled Data

Labeled data is a group of samples that have been tagged with more than one label. Labeling typically takes a set of unlabeled data and augments each piece of unlabeled data with meaningful tags that are informative. Unlabeled data is consists of data that is either taken from nature or created by human to explore the scientific patterns in it. For example, unlabeled data can include photos, audio recordings, videos, news articles, and tweets.

Uses for machine learning

• Image Recognition: Implementing machine learning into image recognition involves making a computer learn, through a learning model, the different key features of an image. Some examples of its uses are pattern recognition, face detection, face recognition and optical character recognition.

• Sentiment Analysis: Implementing machine learning into Sentiment Analysis is a process of determining the opinion or attitude of a speaker or a writer in the tone they use to express themselves.

• News Classification: Using machine learning for News Classification lets us determine what type of content a user is asking to see. It's like telling a computer to give me the sports channel without telling it to give me the sports channel.

• Information Retrieval: It is the process of extracting structured data from the unstructured data.

• Medical Services:  Machine learning methods, tools are used extensively in the area of the medical-related problem. For Example: As an instance to detect a disease, therapy planning, medical-related research, prediction of the disease situation. Using machine learning software in the healthcare brings a breakthrough in medical science.

• Social Media: Using the machine learning approach, social media can create attractive features, giving recommendations, reactions, topics of interest for the users.

• Predictions: Prediction is the process of saying something based on past history. Examples can be weather prediction, traffic prediction, stock market and more. 

Learning Algorithms

Machine learning has several learning methods, it's main methods are: Supervised Learning, Unsupervised learning, Semi-supervised learning and reinforcement. It collects input which is data collected from a source and it produces an output that is the result of the data input. Machine learning models require a lot of data in order for them to do well. In order to train a machine using a learning model, one needs to collect representative sample of data from a training set. Data from the training then can be set in varied like texts, a collection of images, and data collected from individual users of a service.

• Supervised Learning: Algorithms are learned from data provided from humans and both the input and the output data are already known to the machine. It experiments with the input and compares its output with the expected output and is able to know whether or not it needs to modify algorithms to match that same expected output. This way of learning can be useful when a future outcome is expected based on specific patterns from the input data.

• Unsupervised Learning: Algorithms are given raw unlabeled data where the output is unknown. With raw input being provided, the program needs to be able to come up with ways to use it and categorize it. Then, the program studies the input every time and eventually learns how to use it in order to find hidden and complex patterns. 

• Semi-supervised learning: This method falls in between into Unsupervised and Supervised learning, using supervised learning to make better predictions and come up with better patterns. Facial recognition is a good example of Semi-supervised learning. 

• Reinforcement: The program is exposed to many sets of data that and based on that exposure it comes with a solution. Reinforcement learning is dependent since it makes decisions sequentially so all later input data depends on the previous output. In reinforcement learning, we have components called agent and reward. The agent is the decision maker and is supposed to come up with the best way to get to the reward, that is the most desired outcome. Reinforcement is used in robotics, navigation, industrial automation, creation of training systems, and gaming.

Importance of Machine learning

Due to its global demand companies are using machine learning to increase their productivity. Task that would otherwise take most likely an hour to complete are done in seconds thanks to automation. Car factories around the world use robotics with learning modules to perform tasks that would take an entire team to complete. Another example is the email automation, massive amounts of marketing emails are sent daily to user in order to inform them about their products or services and its completely done by a computer. Self-driving cars are becoming more of a reality than science fiction but, we should also see the down side of automating jobs.

Morals of Machine Learning

One of the questions asked about automation is: Is it morally correct for a machine to take the job of a human? The field of AI was created under the assumption that human intelligence could be replicated into a machine, however, there is mixed feelings about this topic. This raises philosophical arguments about the nature of the mind and the ethics of creating artificial minds endowed with human intelligence. AI has been seen as fiction for the most part until recent days, the fact that we can trade human mistakes for efficiency will lead to a risk of mass-unemployment. In the end, Machine learning has been applied to a lot of things in our daily life.

In the end...

Machine learning involves several training techniques that a machine uses to perform tasks in a  similar or superior ways to the human brain. There are a lot of things to take into account when it comes to the performance of a learning machine. These include the type of data that’s being used The methods applied: supervised, unsupervised, semi-supervised. Also, the algorithms that follow those learning methods and the math behind it. The way we incorporate machine learning into our lives is also important and has made our way of life easier. It is as if all the things we see in science fiction are now becoming a reality. Machine learning has its ups and down but it is something that will define the future of technology.

References

1. https://en.wikipedia.org/wiki/Machine_learning
2. https://www.ubuntupit.com/top-20-best-machine-learning-applications-in-real-world/
3. https://www.investopedia.com/terms/d/deep-learning.asp
4. https://en.wikipedia.org/wiki/Artificial_intelligence
5. https://intellipaat.com/community/2809/what-is-the-difference-between-labeled-and-unlabeled-data