1 / 6

Types of machine learning

Our group of industry specialists gives inside and out information on AI online courses including constant tasks, making you a specialist in creating calculations by applying administered and unaided learning, relapse, and time-series models.<br>https://nearlearn.com/courses/ai-and-machine-learning/machine-learning-with-python-training<br><br>

NL1
Download Presentation

Types of machine learning

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Types of machine learning Machine learning for the average person: An analysis of the types of machine learning written for the average person. Machine Learning

  2. At a higher level, machine learning is simply the study of teaching a computer program or algorithm how to progressively improve on a set task given to it. On the research side of things, machine learning can be viewed through the lens of theoretical and mathematical modeling of how the process works. However, more practical is the study of how to build applications that demonstrate this iterative improvement. There are many ways to formulate this idea, but there are broadly three recognized categories: supervised learning, unsupervised learning, and reinforcement learning. In a world full of over-enthusiastic talk about artificial intelligence, machine learning, and both, it is interesting to learn to understand and identify the types of machine learning we face. For the average computer user, this can take the form of understanding the types of machine learning and how they manifest themselves in the applications we use. And for practitioners creating these applications, it is essential to know the types of machine learning so that for any task you may be faced with, you can create the appropriate learning environment and understand what made you work. supervised learning Supervised learning is the most popular paradigm for machine learning. It is the easiest to understand and the easiest to implement. It is similar to teaching a child using flashcards. Given the data in the form of examples with labels, we can feed these example-label pairs one by one to the learning algorithm, allowing the algorithm to predict the labels for each example, and it can respond. whether it predicts the correct answer or not. Over time, the algorithm will learn to infer the exact nature of the relationship between instances and their labels. When fully trained, the supervised learning algorithm will be able to see a new, never-before-seen example and predict a good label for it. Because of this supervised learning is often described as task-oriented. It focuses excessively on a singular task, feeding the algorithm more and more examples until it can perform accurately on that task. This is the type of learning you will encounter most, as it is demonstrated in many of the following common applications: Ad Popularity: Selecting well-performing ads is often a supervised learning task. Many of the ads you see when you browse the Internet are placed there because the learning algorithm said they were of reasonable popularity (and clickable). Furthermore, its placement on a certain site or associated with a certain query (if you find yourself using

  3. a search engine) is largely due to a learned algorithm that says the match between the ad and the placement will take effect. Spam Classification: If you use modern email systems, chances are you've encountered spam filters. That spam filter is a supervised learning system. Fed email example and label (spam/not spam), these systems learn how to preemptively filter out a malicious email so that their use is not bothered by them. Many of these also behave in such a way that a user can assign new labels to the system and it learns the user's preference. Face Recognition: Do you use Facebook? Most likely your face has been used in a supervised learning algorithm that has been trained to recognize your face. Having a system that takes a photo, finds faces, and guesses who's in the photo (suggesting a tag) is a supervised process. It has multiple layers, finding and then detecting faces, but it is still monitored. Unsupervised Learning Supervised learning is the exact opposite of supervised learning. It has no label. Instead, our algorithm will be fed a lot of data and given the tools to understand the properties of the data. From there, it can learn to group, cluster, and/or organize the data in such a way that a human (or another intelligent algorithm) can come and understand the newly organized data. What makes non-supervised learning such an interesting field is that the overwhelming majority of data in this world is unlabeled. Having intelligent algorithms that can take our terabytes and terabytes of unlabeled data and decipher it is a huge source of potential profit for many industries. This alone can help increase productivity in many areas. For example, what if we had a huge database of every research paper we've published so far and we had an untrained learning algorithm that knew how to group these in such a way that you were always current in a particular area of research? Be aware of progress. Now, you start undertaking a research project yourself, connecting your work to a network that the algorithm can see. As you write your work and take notes, the algorithm gives you suggestions about related works, works you want to cite, and works that can also help you advance that area of research. With such a tool, your productivity can be enhanced immensely. Because unsupervised learning is based on data and its properties, we can say that unsupervised learning is data-driven. The results of unsupervised learning are governed by the data and the way it is formatted. There are some areas that you can see as unsupervised learning crops:

  4. Recommendation system: If you've ever used YouTube or Netflix, you've probably encountered a video recommendation system. These systems are often placed in unsecured domains. We know things about the videos, probably their length, their style, etc. We also know the viewing history of many users. Considering users who have watched videos similar to yours and then enjoyed other videos that you haven't seen yet, a recommendation system can look for this relationship in the data and prompt you with such a suggestion. can give Buying Habits: It is likely that your buying habits are contained somewhere in a database and data is being actively bought and sold at the moment. These buying habits can be used in unsupervised learning algorithms to group customers into similar purchasing segments. It helps companies market to these grouped segments and may even be similar to a recommendation system. Grouping User Logs: Less user-facing, but still very relevant, we can use unsupervised learning to group user logs and issues. It can help companies identify central themes to problems facing their customers and rectify these issues, either by improving a product or by designing FAQs to handle common issues. Either way, it's something that's actively done and if you've ever submitted an issue with a product or submitted a bug report, it's possible that there is a way to cluster it with other similar issues. was fed to an unsupervised learning algorithm. Reinforcement Learning Reinforcement learning is quite different as compared to supervised and unsupervised learning. While we can easily see the relationship between supervised and unsupervised (the presence or absence of labels), the relation to reinforcement learning is a bit ambiguous. Some try to tie reinforcement learning closer to the two by describing it as a type of learning that relies on a time-dependent sequence of labels, however, my opinion is that it makes things more confusing. I like to see reinforcement learning as learning from mistakes. Put the reinforcement learning algorithm in any environment and it will make a lot of mistakes in the beginning. As long as we provide the algorithm with some sort of signal that associates good behavior with a positive signal and bad behavior with a negative one, we can fine-tune our algorithm to prioritize good behavior over bad ones. Over time, our learning algorithm learns to make fewer mistakes than before. Reinforcement learning is very behavior-driven. It has influences from the fields of neuroscience and psychology. If you've heard of Pavlov's dog, you may already be familiar with the idea of a reinforcing agent, even if it's biological.

  5. However, to really understand reinforcement learning, let's break down a concrete example. Let's take a look at teaching an agent to play Mario games. For any reinforcement learning problem, we need an agent and an environment as well as a way to connect both through a feedback loop. To connect the agent to the environment, we give it a set of actions that can affect the environment. To connect the environment to the agent, we have the agent issue two consecutive signals: an updated state and a reward (our reinforcement signal for the behavior). In a game of Mario, our agent is our learning algorithm and our environment is the game (possibly a specific level). Our agent has a set of functions. These will be our button states. With the passage of time, our updated stat will be each game frame and our reward signal score will change. As long as we tie all these components together, we will be setting up a reinforcement learning scenario for playing Mario games. Where is reinforcement learning in the real world? Video Games: One of the most common places to look for reinforcement learning is learning to play games. Check out AlphaZero, and AlphaGo, Google's reinforcement learning applications, which learned to play Go games. Our Mario example is also a common example. Currently, I don't know of any production-grade game that has a reinforcement learning agent deployed as its game AI, but I can imagine that it will soon be an interesting option for game developers to employ. Industrial Simulation: For many robotic applications (think assembly lines), it is useful that our machines learn to accomplish their tasks without hardcoding their processes. It can be a cheap and safe option; It may also be less prone to failure. We can also encourage our machines to use less power, so as to save our money. Plus, we can start all of this within a simulation so that we don't waste money if we potentially break our machine. Resource Management: Reinforcement learning is good for navigating complex environments. It can handle the need to balance certain requirements. Take Google's data centers, for example. They used reinforcement learning to balance the need to meet our power requirements, but did it as efficiently as possible, resulting in huge cost reductions. How does this affect us and the average person? Affordable data storage costs for us as well and less impact on the environment than we all share. TYING ALL TOGETHER Now that we have discussed the three different categories of machine learning, it is important to note that at times the lines between these types of learning become

  6. blurred. Furthermore, there are many tasks that can easily be defined as one type of learning and then converted to another paradigm.

More Related