Introduction to Machine Learning

 

Introduction to

Machine Learning

Read also :  https://datasciencemarketing.blogspot.com/2022/04/que-est-ce-que-cest-un-arbre-de-decision.html

In the recent years, machine learning has become a popular subject. It plays such an important role in our daily life and it is present in so many different fields that we do not even realize it when we use it. A personalized product recommended by Amazon, language translation with Google Translate, information research with Safari, spam mails detected by Hotmail, virtual assistance Siri… All of these are common applications of machine learning in our daily life.

Machine learning is part of artificial intelligence, which aims to build systems that can learn or improve performance based on the data used. There are three types of machine learning, which are supervised learning, unsupervised learning, and reinforcement learning. Supervised learning needs to have clear goals and know exactly what results we want. Models are trained on known input and output data so that machine can predict future unknown outputs. Image classification is a typical example of supervised learning, where machines must answer questions like “Is the image a cat or a dog?”. Being different from supervised learning, unsupervised learning does not require a complete set of input and output data, which means machines need to find their own way without any guidance. It is mainly used to explore hidden patterns and distributions. Reinforcement learning emphasizes how to act based on the environment to maximize the expected benefits. Machines will not be told what actions should be taken but must try to discover which actions will produce the most profitable.

There are seven steps in the whole process of machine learning. Once the problem is defined, the first step of machine learning comes to collect data. The quantity and quality of the data directly determine the quality of the prediction model. After gathering data, we will put all our data together and then randomize the ordering of the data since we do not want the order of the data affect what machine learns. We will also put our data in two parts, one training part and one evaluating part. Then we need to choose a model according to our data types. The next step is the training step where we use our data to improve gradually our model’s ability to predict. After that comes the evaluation step where we will use the second part of the dataset to test our model and to see how it performs. Before the final step, parameters which are set at the beginning needed to be tuned to improve our training. And finally, machines are able to predict and to answer previous questions. It is at this point that the value of machine learning is realized.

There is no doubt that machine learning is a powerful tool which is widely used in many sectors. It not only helps us deal with huge amount and various types of data but also allows us to easily identify the trends and patterns without too much human intervention. Moreover, it keeps improving in accuracy and efficiency while learning. However, sometimes we may have a problem to acquire data since it demands huge datasets of goof quality. The ability to interpret the results generated by the algorithms may also be a challenge. What is worse, if we use a biased dataset for training, it is probably that we get biased predictions, which will mislead users to make wrong decisions. 

The process of machine learning is just like a new child learning from itself. It enables companies to discover new trends from large and diverse datasets without too much effort. What is more, it allows companies to deliver new, personalized, or distinctive products and services. Therefore, considering machine learning as a strategic initiative can be a profitable decision. However, we should be really careful while choosing the dataset as well as the algorithms, or this might lead us to a totally wrong way.