Machine Learning

Nagpur Institute of Technology

Machine Learning


Machine Learning

While shopping online for products, have you ever noticed the recommendations you get for a product similar to what you’re looking for? For instance, while shopping at Amazon, you must have noticed how the site recommends, “customer who bought this product also bought this,” while referring to a combination of products. How are they making these recommendations? Well! This is Machine Learning.

Machine Learning

While shopping online for products, have you ever noticed the recommendations you get for a product similar to what you’re looking for? For instance, while shopping at Amazon, you must have noticed how the site recommends, “customer who bought this product also bought this,” while referring to a combination of products. How are they making these recommendations? Well! This is Machine Learning.

What is Machine Learning (ML) and What Does it Do?

Machine Learning is a concept that allows the machine / computer to learn from examples, instances and experiences, without human intervention or assistance.  Machine Learning (ML) focuses primarily, on the development of computer programs that can access data and use it for themselves.  The process of learning begins with accumulating the data, such as instructions or examples, and in the process, a specific pattern is established, which is used for making future decisions or predictions based on those examples provided.  Now, this accumulated data enables the computers or machines to make data-driven decisions carry out a certain task. These programs or algorithms (as they are called) are designed in a way that they learn and improvise over time whenever they are exposed to a new set of data. The main objective is to allow the computers to learn automatically without any human intervention and adjust the actions accordingly.

How Does Machine Learning (ML) Work?

Machine Learning algorithm is devised using a training data set to create a model. When a new input data is introduced to the Machine Learning algorithm, it makes a prediction on the basis of that model. This prediction is then evaluated for accuracy and once the accuracy is established, the Machine learning algorithm is deployed. On the other hand, if the accuracy is not confirmed, the Machine Learning algorithm is trained again with an improved and more enhanced training data set.

Machine Learning uses two types of techniques:

  1. Supervised Learning- which trains a model on the known input and output data so that it can predict the future outputs. In other words, supervised learning is where you can consider as if the learning is guided and supervised by a teacher. The data set acts as a teacher and its role is to train the model or the machine. Once the model or the machine is trained, it can start making predictions and decisions based on the new data given to it.
  1. Unsupervised Learning- which finds hidden or disguised patterns or some intrinsic structures in input data. The model or the machine in unsupervised learning learns through observation and finds structure in the data. Once the model is provided with a data set, it automatically finds patterns and relationships in the data set by creating clusters in it. The limitation, however, is that it cannot add labels to the cluster. (Clustering is the most common unsupervised learning technique. It is used for exploratory data analysis or EDA to find hidden or disguised patterns or groupings in the data) For instance, it cannot say this is a group of ice-creams and chocolates, but it will separate all the ice-creams from the chocolates.

Key Takeaways in Applied Machine Learning

  • The most important factor in Machine Learning (ML) is the domain specific data features and how to train your models or machines based on the adequate data accumulated.
  • One of the limitations of Machine Learning (ML) is when algorithms don’t perform well. This is primarily due to a problem with the training data. In other words, insufficient and skewed data or insufficient features describing the data for making predictions or decisions  often leads to the non-performance of an algorithm. 
  • Since, you would want your learning algorithm to perform well on fresh data, it is imperative to set aside a chunk of your training data set for cross validation.

 

Why Machine Learning Matters?

Machine Learning facilitates the analysis of massive quantities of data and is fast becoming a key technique in solving problems in areas such as:

  • Automotive, aerospace and manufacturing industries for predictive maintenance
  • Computational Biology for drug discovery, DNA sequencing and complex disease detection
  • Computational Finance for credit scoring and finance trading based on algorithms
  • Energy Production for electricity load forecasting and its price
  • Image Processing and Computer Vision for face recognition (used for biometric security, image and video indexing systems) and object detection (highly useful in video surveillance and to retrieve images.)


LEAVE A COMMENT

  

Subscribe to our Newsletter for latest news.

© NIT 2017 - All rights reserved. | Powered By BigAdTruck India