Introduction to Machine Learning

These days  wherever you go or whatever you use, there is a slightest hint of "Machine Learning" invloved and you definitely wonder what in the Earth do "Machine Learning" mean. Machine learning is a very interesting topic and once when you're into it, there's no going back, you're gonna just want more of that. Machine learning just exists everywhere,  every recommendation you receive on any popular platform, there has to some machine learning algorithms involved in it.

Actual Definition-

Machine learning is the study of computer algorithms where the machine is trained to improve automatically through experience and from the knowledge of data. In simple words, you instruct the machine with codes to learn something and it can replicate what it learned. It can improvise over data and efficient algorithms. Generally machine learning involves a lot of mathematics and hell load of data.  

Machine learning is a part of our life even without knowing that it even existed in the first place. It is used in most of the virtual assistants like Alexa, Siri and popular platforms like Youtube, Google maps. We will discuss the uses and real life applications later in the blog.

Machine learning is considered as an important technology, since automation gets priority before anything else and machine learning is the one you turn to. If you work for an Artificial intelligence project, there's no way that the project can exist without Machine learning. It plays a significant role in Artificial Intelligence and AI plays a major role in several places these days.

How it works? So, a machine learning model gets trained over a set of data called "Training set". It can do its own predictions or decisions without even being programmed to do so. The more data is in Training set, the better the model can train and predict new output values.

Theory-

Machine learning is a subdomain under Artificial Intelligence and it allows application softwares to become more accurate so it can predict better outcomes for various purposes. Now, these insights from the data can boost up applications and businesses which results in increase in growth. As I previously mentioned, the machine learning model is trained over the training set. The decision process happens and the model finds the pattern in the data in training set. After the training is done, we can have a set called "Test set". This set contains data which cannot be found in the training set but belongs to the same parent data. We can instruct our model to predict the output values of the test set. But the thing is, the output values of test set is already present and it is compared with the output values predicted by the model. From this comparision, the accuracy and the efficiency is checked. After the error and accuracy check, the model's algorithm can be modified or changed inorder to increase the accuracy of the result. Now the model not only predicts better, it is also optimized to perform better. The three steps "Decision process", "Error Check" and "Model optimization" are the main functions that happen in machine learning.

Uses-

Machine learning is used in various real world applications and here are some :

• Speech Recognition -

            We have seen a mic symbol in various search engines and applications which says "Search by voice" or "Voice to text". It comes under speech recognition and it is a process of converting voice instructions to text. The model is trained on datasets which includes several voice notes and it acts accordingly. Assistants like Alexa, Siri, Cortana use this speech recognition mechanism to process the voice instructions and act accordingly.  

• Image Recognition-

            This subtopic concerns about the machine being taught and able to understand both audio and video. Obviously the machine has to process data as input and then the machine can recognize specific objects like cars, trees, traffic lights, people, rocks. Speaking of cars, Image recognition is the one mainly used in Autopilot softwares to detect objects while driving. "Self- Driving" cars is a great outcome of what machine learning can do.

• Medical Diagnosis-

            With the help of machine learning, prediction of various diseases including cancer, diabetes can be done easily. Medical diagnosis is one of the process in hospitals which can be very time consuming with the existing traditional methods. Thanks to ML, now medical diagnosis can be done faster than ever. ML has demonstrated truly life-impacting potential in Medical Diagnosis which can help to save lives.

• Recommendations-

            As I mentioned before, ML is being used by various popular entertainment and E-Commerce companies such as Amazon, Youtube, Netflix, Google for personal recommendations. Once you search for a video in youtube, you're going to receive recommendation of video on the same type of video you watched earlier. Youtube learns what we watch, since the Youtube history is turned on and it suggests videos. If you search for a product anywhere, the model in the application learns that and will send you suggestions of the same product wherever you go in the internet. 

Once, I turned off the Youtube History which includes "The history of the videos I watched" and "Youtube search history". This must have disturbed the Youtube algorithm somehow that youtube started to show me the videos I have watched already in my home page and shorts. I never got any new videos or shorts until I turned on the Youtube History. This is something that Google and Youtube team should work on. User's privacy should be given the most priority and if the user wishes to turn off the history, the platform still needs to perform well for that user too.

Machine learning can be utilized to solve real-world problems and I believe it can truly revolutionize the current technology and change our way of living life with the help of technology. There are so many algorithms used, sub-domains and different types in Machine Learning. That's for the next upcoming blogs.