What is Machine learning | Machine learning algorithms | Total Tech

Machine learning

What is Machine Learning?

Machine Learning is the learning of machines where they do all things that humans can do like a prediction of rain if the atmosphere is cloudy. In the future, they can do more good predictions as compared to humans.

Machine learning is the most demanded thing in the world because Machine Learning is the branch of AI that can connect humans with machines. Its demand is increasing day by day. It is like the training of a computer. In that training, we prepare our computer to solve any task.

We can train our machine by giving some data related to your task. After your machine has read your data then make a data pattern then give predictions according to your data. Your machine didn’t give exact results but it gives accurate results.

Machine learning is a subset of Artificial Intelligence. In artificial intelligence, your machine was dependent on himself but in machine learning your machine was dependent on your data.

Machine learning have three types of categories:

• Supervised learning
• Unsupervised learning
• Reinforcement learning

To learn machine learning we have to know:

• Mathematics.
• Programming skills.
• MySQL and NoSQL Database.
• Any programming language like python, R.
• Data Preprocessing and Data Visualization.
• Spend time on machine learning and daily practice.

Now, You follow all these steps then improve yourself day by day. You can also join some courses online.

Machine learning applications

Recommendation system- In Netflix, Amazon, Youtube, Google, etc. They all use a recommendation system When any customer comes so this system recommends buying the best products or Video. When a user searches any on those platforms, At the top Position they place that product where a user has an interest. 

They also show you that advertisement on that product you search for. That’s the reason that people prefer google products because this recommendation algorithm is more powerful, user-friendly, Trusted. Their recommendation system is explicitly programmed and powerful. A recommendation system was created with the help of machine learning.

Speech Recognition- In many applications like Google, Alexa, Youtube, Microsoft, you see there is an option to speak and they converted our voice to text this was happening with machine learning. This speech recognition first learns the voice and then predicts the text according to the voice, after the voice change in a text then they give such a result that the user wants this was made more user interaction. 

The speech recognition was mostly useful for 50+ old people who were not able to type on keyboards and This is also used by those people who did not have such knowledge about technology.

Image Recognition- In Image recognition you upload any image on the computer then it will detect this Image then It will similar images on its database then it will provide you the similar images about your image. This Image recognition was happening with computer vision technology.

Let’s see one Image recognition application In the Facebook application you notice whenever you upload your image, then Its automatically show your friends and your relatives profile this was using image recognition Facebook takes your image then it will see all the related images in his database then he see the picture where you and your friends both in that picture then it finds your friends profile then it will show you.  

Self Driving Cars- With the help of machine learning the self-driving cars were possible with this technology you can automate your driving. Your car automatically drops at your destination. This technology is most useful for people who drive too long and they have to take some rest. 

If you did not know about driving so this technology is not useful for you because If your computer was not working properly or had any problem, so, Unfortunately, It will have an accident. So, This technology is totally in your hands whether it is useful or not.

This Technology has not come. But Tesla cars work on this self-driving technology. This technology came with a tesla car, then in the future, it can also coming With other cars.

Machine-learning introduction:

What is Supervised learning?

In supervised learning, we give some input data then your machine is giving predictions as output data. Machine learning has two terms Features and Labels. Features are the input data that we give to our computer. Labels are the data predicted by the computer as output data or labels. 

Let’s take an example- If you want to classify the dogs and the cats so, You can classify the dog by its eyes, nose, ears, tail, voice, etc. After seeing all these factors on any animal you can say this is Dog and you can classify the cat by its color, hair, body structure, height, face, voice, tail, etc. 

After seeing all these factors on any animal you can say this is Cat. When any other factor is given to our computer so it can easily predict It is a dog or a cat. The factors of the animals are featured and the result that you say this is a dog or cat is labeled.

Supervised learning is of two types:

• Classification
• Regression

Classification- In classification, you classify the data. In a simple way to understand classification, The data that your computer predicts comes in a label category like in the above example you classify the dogs and cats by their factors. In that example, you know that it was predicted only by a dog or cat. It does not predict any other animal because we do not mention any other animal factors and its name.

Regression- In regression, you predict any number. That number you predict is not important to mention on data for example if you want to give a percentage of students out of 100, So you create data on the performance of the student, behavior of the student, Knowledge of the student, marks of the student. After seeing all the data you can predict any number according to his data. 

If we have a small amount of labeled data and featured data, our machine learning model did not give more accurate results because our machine was learning from data. In that case, we use semi-supervised learning.

Semi-supervised learning was included in machine learning. Semi-supervised learning was created for a small amount of data. When we learn machine learning this concept comes between unsupervised learning and supervised learning.

What is Unsupervised learning?

In Unsupervised learning, we did not have any labeled data, we have only features or input data. We train our model without labeled data; our model uses pattern recognition to learn from data. Our model was to learn the hidden patterns of data, the Unsupervised learning model was to learn from the past by itself.

To understand more properly, let’s take one real-world example- when any child was born he learns from his parents. His parents taught him how to eat, how to walk, etc. After some years, the child was going to a school where he sees how his friends handle any situation, he sees how his friends play the games, in that case, he learns by himself he didn’t need help from his parents.

In unsupervised learning, the same thing happening in our model was learned with features, he did not need the help of the labels. Unsupervised learning helps to detect any genetic disease by the blood cells, etc.

Unsupervised learning has One type:

• Clustering

Clustering- It is the type of machine learning where it does the training of the data without giving output data. This using neural networks. It justifies the data by Trial and error.

To Understand clustering let's take an example- Suppose you have 5 animals Dog, Cat, Lion, Tigar, and Monkey. If you have an image of a Lion and you don’t know how this animal looks. You only know the name of these animals, so After seeing this image You First Guess this is Dog, Second Guess this is Cat, and Third Guess this is Lion. 

In that case, you will guess these animals' names again and again, After your guess goes to the right one you stop it, and next time you give the same picture so you didn’t have to guess it because you already guessed the picture of the lion. The same thing happens in clustering.

How scikit learn helps in machine learning?

Scikit learn is the library of python that helps in machine learning. Scikit learn has many algorithms for machine learning with the help of Scikit learn you can easily create any model by using its algorithms. 

Scikit learns consists of all the tools and algorithms that we get help in machine learning. With the help of these tools, we can do data splitting, data shuffling, handle categorical data, etc.

After Machine learning, we do deep learning. In deep learning, we use the TensorFlow library of python. This library was created by google. In deep learning scikit learn does not help us.

Some Important libraries for machine learning:

• Scikit learns- It consists of all algorithms and tools that help in the machine learning model. This creates a more accurate model.

• Numpy- This helps in numerical operations work with next-level mathematical operations and it also helps in matrices and arrays.

• Pandas- Pandas is the most popular library for data preprocessing. It helps in data analysis, data cleaning, etc. It is the most powerful library of python.

• Matplotlib- It helps in data visualization because if we have to understand our data more deep we use matplotlib for data visualization. If we have to show anything from our research, we didn’t show the code we show the data visualization.

• Seaborn- It is advanced level data visualization it gives more rights for data visualization. It is similar to matplotlib.

• Tensorflow- It is for Deep Learning this helps in deep neural networks. This library was created by google. This is used when we finish machine learning, then start Deep learning.

Python consists of more alternative and different libraries for machine learning, But most of the time these libraries helped you and the other libraries were used according to your machine learning problem. For data analysis, you have to learn big data for that you have to learn NumPy, Pandas, Matplotlib. These three libraries for data analysis. 

Machine learning algorithms:

Machine Learning has many algorithms for each of these categories you have to learn these all algorithms. Let’s see all those popular and Best algorithms.

Machine-learning algorithms for Classification.

• Logistic Regression.
• Decision Tree.
• Support Vector Machines(SVM)
• Naive Bayes
• Nearest Neighbor
• Discriminant analysis
• Random Forest
• Stochastic Gradient descent

Machine-learning algorithms for Regression.

• Linear Regression
• Support Vector Regression(SVR)
• Neural Networks
• Decision Tree
• Lasso Regression
• Logistic Regression
• Multiple Regression algorithm
• Multivariate Regression algorithm

Machine-learning algorithms for Clustering.

• K-Means
• Fuzzy C-Means
• Hidden Markov Model
• Mini Batch
• Mean Shift Clustering
• Agglomerative Hierarchical Clustering
• Hebbian Learning
• DBSCAN

If you really feel that you get enough knowledge for the base machine learning then for whom you wait let's start machine learning today and if you have any queries then comment me.