This article gives you a clear understanding of the Machine Learning concept for beginners. So let’s start our Machine Learning Tutorial for Begnners blog post.
History of Machine Learning
What is Machine Learning?
- Machine Learning is a science that allows the machine to learn from experience without being explicitly programmed. Or in other words, It’s the art of getting computers to learn and act as humans do.
What are Machine Learning pre-requisites?
Machine Learning Pre-requisites for Beginners are:
- Knowledge of probability and linear algebra.
- Knowledge of any computer programming language. Good to know Python programming language.
- Knowledge of Calculus, especially derivatives of a single variable and multivariate functions.
How does Machine Learning work?
Like a Brain!
Yes, Let me explain in simple terms,
If a “ model” is exposed to any image or an object, it will “predict” what the image/object is all about.
Here model and predict means:
- Model – is what we call a Machine Learning algorithm which is not coded i.e if the image is round and in orange color then it’s Orange fruit.
- Predict – As the name itself tells you that, the model will start predicting the objects, images, etc…which helps in storing the feedback. ( Image below)
In Technical terms
- When new input is given to the ML model, it analyzes the data and applies its learned patterns over the new data to make future predictions.
- Based on the final prediction, the model stores the feedback and can optimize the model using various standardized approaches.
- This approach is adopted in Machine Learning models.
Types of Machine Learning Algorithms
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
- This is a Task-Driven algorithm that uses training data and feedback from humans to learn the relationship of given inputs to a given output.
There are two categories of supervised learning:
- Classification task
- Regression task
|Linear regression||Find a way to predict future values.||Regression|
|Logistic regression||Logistic regression is a classification task and extension of linear regression.||Classification|
|Decision tree||A decision tree is a regression model that splits data values into branches at decision nodes.||Regression Classification|
|Naive Bayes||Naive Bayes makes use of the Bayesian theorem. The theorem helps in getting to know the event prior.||Regression Classification|
|Support vector machine||Support Vector Machine algorithm is a non-linear solver that finds a hyperplane that optimally divides the classes and this is typically used for the classification task.||Regression (not very common) Classification|
|Random forest||Random forest algorithms help in improving the accuracy which is built upon decision trees.||Regression Classification|
|Gradient-boosting trees||Gradient-boosting trees help in focusing on the error committed by the previous trees and try to correct it and this is also a classification technique.||Regression Classification|
- This is a Data-Driven algorithm, this helps you when you do not know how to classify the data and you want the algorithm to find patterns and classify the data for you
- Examples of Unsupervised Learning: Apriori algorithm, K-means.
|K-means clustering||Clustering puts data into some groups that each contain data with similar characteristics.||Clustering|
|Gaussian mixture model||Gaussian mixture model provides more flexibility in the size and shape of groups||Clustering|
|Hierarchical clustering||Hierarchical clustering can be used for loyalty-card customers which helps the splits along a hierarchical tree to form a classification system.||Clustering|
|Recommender system||The recommender system helps in defining the relevant data for the recommended system.||Clustering|
|PCA/T-SNE||PCA/T-SNE is used to decrease the dimensionality of the data.||Dimension Reduction|
- Reinforcement Learning is popularly known as the learning for error algorithm and this is one of the most popular types of Machine Learning Algorithm which is used in cars and industrial robotics.
- This algorithm aims to reach a goal while learning from the previous errors in a dynamic environment.
- Example: Markov Decision Process.
Difference between AI, ML deep learning
|Artificial Intelligence||Machine Learning||Deep Learning|
|Artificial Intelligence is the study/process which enables machines to mimic human behavior through particular algorithms.||Machine Learning is the study that uses statistical methods enabling machines to improve with experience.||Deep Learning is the study that makes use of Neural Networks to imitate functionality just like a human brain.|
|AI consisting of ML and DL as it’s components.||ML is the subset of AI.||DL is the subset of ML.|
|AI is a computer algorithm that exhibits intelligence through decision making.||ML is an AI algorithm that allows systems to learn from data.||DL is an ML algorithm that uses deep(more than one layer) neural networks to analyze data and provide output accordingly.|
|The aim is to increase the chances of success and not accuracy.||The aim is to increase accuracy, not caring much about the success ratio.||It attains the highest rank in terms of accuracy when it is trained with large amounts of data.|
|Three categories of AI are: Artificial Narrow Intelligence, Artificial General Intelligence, and Artificial Super Intelligence||Three broad categories/types Of ML are: Supervised Learning, Unsupervised Learning and Reinforcement Learning||DL can be considered as neural networks with a large number of parameters layers lying in one of the four fundamental network architectures: Unsupervised Pre-trained Networks, Convolutional Neural Networks, Recurrent Neural Networks, and Recursive Neural Networks|
|The efficiency Of AI is the efficiency provided by ML and DL respectively.||Less efficient than DL as it can’t work for longer dimensions or higher amounts of data.||More efficient than ML as it can easily work for larger sets of data.|
|Examples: Google’s AI-Powered Predictions, Ridesharing Apps Like Uber and Lyft, etc.||Examples: Virtual Assistants: Siri, Alexa, Google, etc.,||Examples: Sentiment based news aggregation, Image analysis, and caption generation, etc.|
Why is Python a good choice for Machine Learning?
Here are some of the reasons :
- An extensive selection of libraries and frameworks
- The simplicity: Python is Easy To Use
- Python has Community and Corporate Support
- Python is Portable and Extensible
- A low entry barrier
- Platform independence
- Good visualization options
Machine Learning Applications
- Google’s Map: Google Maps provide insights like traffic, construction roads, accidents by using the location data from smartphones. This is Machine Learning.
- Riding Apps: Riding Apps like Uber OLA uses Al and ML to assist the user with the estimated price of a ride, optimal pickup location, and ensuring the shortest route of the trip.
- Spam Filters: Gmail uses AL and ML to filter Spam mail.
- Smart Replies: Gmail prompts simple phrases to respond to emails like “Thank You”, “Noted”, “Yes, I’m interested”. These responses are customized with the help of ML and AI.
Evaluation and Assessment
- Plagiarism Checking: ML can be used to build a plagiarism detector.
- Robo-readers: With the help of AL and Ml algorithms “reading “ can be done by robots. Eg: the GRE uses a Robot called “ e-Rater”.
- Facebook: Friends suggestions, automatically reflecting friends’ faces is all because of AL, ML. Facebook uses AI and ML to identify faces.
- Instagram: With the help of ML algorithms, Instagram can make and auto-recommend emojis and emojis hashtags.
Medical Diagnosis and Healthcare
ML algorithms are highly used for:
- Handling inappropriate data.
- Explaining data generated by medical units.
- Also for effective monitoring of patients.
Machine Learning Limitations
- Ethics: Data mining has resulted in the collection of massive amounts of data, especially by large companies such as Facebook and Google. This is the highest level of privacy breach for the users which is unethical.
- Data – Lack of Data: If you feed a model with poor data, then the result will also be poor. This can happen because of two reasons: lack of data, and lack of good data.
So, Why should freshers learn Machine Learning?
Machine Learning is one of the important solutions for most of today’s problems be it a top MNC or a startup everyone prefers to go with Artificial Intelligence and Machine Learning. So, if you are a person who wishes to solve the problem and learn more about Machine Learning, check out our free PG program in machine learning that can help you to make your foundations much stronger.