Machine Learning in computer system
What is Machine Learning?
Machine Learning is the concept where machine pretends to do the task as like human does. It’s a broad concept in computer science which does the work of the human tasks in an automatic manner. NLP (Natural Language Processing) is an area of machine learning which is capable to learn, understand, analyze, predict the words of human language. In real life, NLP is used in the field such as Information Retrieval, Information Extraction, Machine Translation, Text Simplification, Sentiment Analysis, Text Summarization, Spam Filter, Auto – Predict, AutoCorrect, Speech Recognition, Question Answering, Natural Language Generation.
Machine Learning is a buzz word in current era. It is widely used in the areas like – Email Spam Filtering, Speech Recognition, Object Detection, Auto Speck Checker in Word, Automatic Self-Driving Car, Recognizing the digits in Number Plate in Vehicles, Catching the objects in traffic signal for breaking protocols.
Difference between AI vs ML vs DL
|Sr. No.||Artificial Intelligence (AI)||Machine Learning (ML)||Deep Learning (DL)|
|1.||AI is originated in 1950s.||ML is originated in 1960s||DL is originated in 1970s|
|2.||AI builds the computer system which can act or mimics like human behavior.||ML is system which can take decisions based on their past experiences.||DL is system which works like humans using artificial neural network (ANN).|
|3.||AI is Superset of ML & DL||ML is a Subset of AI||DL is a Subset of ML|
|4.||AI is much faster as compare to ML & DL||ML is faster as compare to DL||DL is 10 times faster as compare to ML|
|5.||AI has more higher accuracy as compare to ML & DL||ML has less accuracy as compare to AI and DL||DL has higher accuracy as compare to AI & ML|
|6.||AI wants more RAM as compare to ML & DL||ML wants less RAM as compare to AI & DL||DL wants extremely more RAM as compare to AI & ML|
|7.||AI wants more GPU as compare to ML & DL||ML wants less GPU as compare to AI & DL||DL wants extremely more GPU as compare to AI & ML|
|8.||Ex – Alexa Echo, Apple Siri||Ex – Email Spam Filtering, Speech Recognition||Ex – Automatic Recognition of Patterns in E-Commerce websites.|
How does the Machine Learning work?
Machine Learning is the concept in which machines can learn, predict and improves from his past experiences. Below is the architecture of the machine learning cycle.
Let us take an example to get the concept of machine learning, here, we are going introduce NLP which stands for Natural Language Processing, where we are going to filter from the given emails using NLP in ML among the emails which are Spam or not?
What Is Spam? Spam is a type of unwanted messages containing marketing messages often sent to multiple people using bots. Spam email contains contents such as – Advertisement, Sales, Investment, request for banking account, or to order an item, sender’s email having a bunch of numbers succeeding the ‘@’ symbol. Subject lines are not in a formal pattern and would try to attract readers, some may also try to lead readers to phishing sites and various other cyber threats can traverse using these types of messages.
Following are the steps involved for identifying a given email as a spam or a ham?
- We need some packages to done with NLP and need to be install and downloaded.
- For further procedures, we need to collect the datasets to perform operations.
- The computer doesn’t understand human language so that we need to make sure that the computer understands it in their language for that we need to remove the white spaces, full stops, commas, in the sentences, paragraphs.
- After that, we need to clean the given data using common pre-processing techniques so that the unwanted noisy data or irrelevant data must be eliminated from the datasets.
- The First step is removing the punctuation from the given data.
6. Afterward, convert the given the data into tokens
Types of Machine Learning?
There are 3 types of Machine Learning are as follows
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
- Supervised Learning: –
Supervised Learning is a technique in which we can teach the machine using labeled data, under guidance or supervision. Basically, it deals with two types of problem techniques like Classification and Regression. It has a very well-defined training phase so that it can predict the forecast outcomes. It has a direct feedback mechanism because of its known labeled data. It uses algorithms like Linear Regression, Logistic Regression, Support Vector Machine, K- Nearest Neighbour, and Random Forest. Applications such as Risk Evaluation, Forecast Sales, and so on.
- Unsupervised Learning: –
Unsupervised Learning is a technique, in which machines can learn through unlabelled data without any guidance or supervision. Generally, it deals with two types of problem techniques like Association and Clustering. It discovers the patterns based on the data and then grouping by their similarity. It has no any direct feedback because it is unaware of its output data in the training phase. It uses algorithms like K-Means, Apriori, and C-Means. Applications such as Recommendation Systems, Anomaly Detection, and so on.
- Reinforcement Learning: –
Reinforcement learning is a technique in which machines can learn through their own mistakes and later in the future it won’t repeat the same mistakes again. In short in this technique machine can act as an agent, and based on his actions it gets an award or reward. Here, not given any predefined data to the machine. It has both training and testing data. It uses a trial and error method. It has a feedback in the form of reward and punishment from the environment. It uses an algorithm like Q – Learning and SARSA (The State Action Reward and the State Action Algorithm). Application such as Self Driving Cars and building Gaming and so on.