A-A+

The complexity of machine learning in data science

2019-04-30 Education No comment

Machine learning as an API

Machine learning is no longer limited to geeks. Today, any programmer can call some APIs and use them as part of their work. With Amazon Cloud, Google Cloud Platform [GCP] and more, in the next few days, we can easily see that the machine learning model will now be available to you as an API. So all you have to do is process your data, clean up the data and make it into a format that you can finally enter into a learning learning algorithm, which is just an API. So, it becomes plug and play. You insert the data into the API call, the API returns to the computer, it returns the predicted result, and then you take action based on this.

Machine learning – some use cases

Such as facial recognition, speech recognition, recognition of files as viruses, or prediction of today's and tomorrow's weather, all of which can be achieved in this mechanism. But obviously, someone has done a lot of work to make sure these APIs are available. For example, if we do face recognition, there will be a lot of work in the field of image processing, where you can take images, train your model on the image, and finally get a very good image processing. The generalized model can handle some new types of data that will appear in the future, and these data are not used to train the model. This is usually how the machine learning model is built.

Antivirus software case

All anti-virus software is usually identified as malicious or good, benign or secure files. Most anti-viruses have now moved from virus-based static signatures to dynamic machine learning-based detection to identify viruses. Therefore, as antivirus software is used more and more, you will know that most antivirus software will provide you with updates and updates in the past days that have been used for virus signatures. But now these signatures are converted into machine learning models. When there is a new virus update, you need to retrain the models you already have. You need to retrain the model to see if this is a new virus in the market and on the machine. How machine learning can do this is that each malware or virus file has a specific hit associated with it. For example, a Trojan might enter your machine, and the first thing it needs to do is create a hidden folder. The second thing it does is to copy some dlls. The moment a malicious program begins to perform certain operations on your computer, it leaves its orbit, which helps to find them.

leave me a message

Copyright@Springever inc. © Spring All rights reserved.  

User login ⁄ Register

Share: