Computers have grown a lot in computational power since the 70’s. Computers are becoming more capable day by day. One of Intel’s founders Gordon Moore had also stated that the number of transistors on a dense integrated circuit doubles every two years. This has been referred to as the Moore’s Law. This law just shows that the computational power increases with time. At the present stage this observation is no longer applicable as we are not seeing a massive increase in transistor count, which arguably means that computational power is getting close to its peak. This suggests that there is a certain limit on how much computation a computer can handle within a reasonable amount of time. There are certain problems that are too difficult for a traditional computer to solve with hard-coded instructions, like for example predicting if a patient has cancer, if the patient has a chance of getting coronary heart disease etc. These kinds of problems have multiple dimensions and multiple outcomes, mainly probabilistic. There are also certain cases when a developer knows the input and the desired output. In this case if Machine Learning is used correctly, the computer can map the connections between the input and output all by itself. This saves a lot of time for the Developer. Machine learning can also help the Scientific Community and also various indie scientists in predicting the outcomes of certain tests that they want to conduct.
Machine Learning encloses a lot of topics from Mathematics, Physics, Statistics and Geometry. Machine Learning models are classified into two categories: Supervised & Unsupervised Machine Learning.
Under Supervised Learning we have:
- K-Nearest Neighbors
- Support Vector Machines ( SVM )
- Decision Tree
- Naive Bayes
- Random Forest
Supervised Deep Learning Algorithms:
- Artificial Neural Network
- Convolutional Neural Network
- Recurrent Neural Networks
Under Unsupervised Learning we have:
- Clustering Using the K-Means algorithm
- Hierarchical Clustering
- Mixture Models
Unsupervised Deep Learning algorithms:
- Self Organizing Maps
- Boltzmann Machines
- Restricted Boltzmann Machines
- Auto Encoders
Now lets see some of the practical applications of Machine learning in daily life:
- Computer Vision, this is the ability given to a computer to understand what is being shown to it.
- Virtual Personal Assistants, Virtual Assistants such as Google Assistant, Microsoft’s Cortana, Amazon’s Alexa and Siri from Apple are all based on Machine Learning.
- Predictions while commuting, Google Maps uses this technique along with GPS to predict the traffic status of a place.
- Personalized Social Media Content, Using Machine Learning based on the pictures you have liked in the past, Social Sites such as YouTube, Facebook, Instagram, Twitter etc. present pictures or advertisements which you may find interesting
- Spam and Malware Filtering, e-mails generally contain a lot of spam messages as well as viruses using Machine Learning these kinds of data can be filtered out
- Chat-bots, you may have seen in many places such as Amazon and other sites you can get to chat with a bot for customer support! These bots are trained using Machine learning and Natural language Processing.
- Optimized Search Results, sites like google create personalized advertisements using Google AdSense based on your interests. It also enhances your search experience as you are getting content based on what you like.
- Disease prediction, many machine learning models have been trained by Scientists and Doctors to predict the chances of getting Cancer or Heart Disease. This can give some valuable time for the patient to receive critical treatment.
So guys, according to me these were a few detailed descriptive sentences about why is machine learning necessary, and what are its practical applications. That is all for this blog, see you soon!