Machine learning

1st slide

Good afternoon everyone!

Let me briefly introduce myself. My name is Nurmukhammed from IT-2107. As you all know, today I am going to talk to you about machine learning.

My talk is divided into 4 parts. First I will talk about introduction to machine learning. Then in the second part, I'll show my literature review done by myself. And finally I end up with conclusions and references respectively.

Before we get to the first part, I'd like to ask one warm-up question. As you know, robots are based on people, respectively, they have a gender. But, most of robots belong to one particular gender. What gender do you think robots are? Who thinks that robots are male please raise your hands. Thank you all, now, you can put your hands down. Now please raise your hands who thinks that robots are female. thanks to everyone who raised their hands. The answer is Most Robots Are Female. This may surprise you, but it's true. Because, If you ask Siri, Alexa, or even Cortana, most likely, you will be answered by a pleasant and polite woman’s voice. The reason? Studies show that males and females are more attracted to a woman's voice. 

Now let’s move to first part.

In simple words, Machine Learning is a form of Artificial Intelligence which learns on its own. Machine learning algorithms use historical data as input to predict new output values. In some way, if they learn infinitely, they surpass the human race.

The idea of machine learning belongs to Frank Rosenblatt, and it was originally called perceptron and unlike today's machine learning algorithms, it used both analog and discrete signals. If you ask where machine learning can be used, then my answer to you is everywhere, absolutely everywhere. Techniques based on machine learning have been applied successfully in diverse fields ranging from computer vision, spacecraft engineering to biomedical and medical applications.

If talk about evolution of ML, In the 1990s, the evolution of machine learning made a turn. Driven by the rise of the internet and increase of usable data, machine learning suddenly became very popular.

Training artificial intelligence is an energy-intensive process. New studies show that the carbon footprint of training a single AI is as much as about 300 tonnes of carbon dioxide, in other words, five times the lifetime emissions of an average car. For more details, you can take a look at the slide.

If we talk about the social effects of machine learning, then they depend on your goal. You can create a new terminator or you can predict people's cancers. If we talk about today, machine learning brings more benefits to society than harm. Why did I choose machine learning over another invention? Because machine learning is trendy thing, and it has the potential to turn the world around, also machine learning specialists get paid a lot. Machine learning algorithms have so many real-world applications, including Walmart's product recommendations, and Facebook, Instagram's content on users' feeds.

Let’s move to Literature Review.

The problem of researching this article is "How we can track and predict the energy consumption of machine learning algorithms?". They used an open source tool called Carbontracker, which available on Github. They found that machine learning is significant contributor to climate change.

The second article gives a brief information about machine learning and describes well-known algorithms.

The third article is called "Early History of Machine Learning" and provides brief information about the history of machine learning.

The fourth article is called "Machine Learning Applications for Data Center Optimization" written by Jim Gao. This article has two research questions: "How we can reduce energy consumption?" and "How we can improve efficiency of machine learning algorithms?". He used the python programming language and its libraries, Scipy and Numpy, also, he used software called Matlab. He discovered that we can use machine learning itself to improve its efficiency.

The latest article is about cancer prediction. The research problem of this article is the lack of available data about cancer. They used the biomedical database scopus as a methodology and found that in the future robots will be able to outperform humans in predicting cancer.

That's my conclusion. Finally, I'd like to say that machine learning revolution will stay with us for long and there will be era of machine learning.

You can take a look at my references. I used exactly 5 articles and here they are.