Shulou(Shulou.com)06/01 Report--

This article mainly introduces the relevant knowledge of the visual representation of machine learning, the content is detailed and easy to understand, the operation is simple and fast, and has a certain reference value. I believe you will gain something after reading this visual representation of machine learning. Let's take a look.

What is machine learning?

"Machine learning is a science that allows computers to learn and act like human beings, improving their learning in an autonomous way by providing them with data and information in the form of observation and interaction with the real world."

The above definition contains the ideal or ultimate goal of machine learning, as expressed by many researchers in this field. The purpose of this article is to provide business-savvy readers with expert views on how to define machine learning and how it works. Machine learning and artificial intelligence have the same definition in many people's minds, but readers should also be aware of some obvious differences. References and interviews with relevant researchers are included at the end of this paper for further mining.

How do we get to our definition: like any concept, the definition of machine learning may be slightly different, depending on your requirements. We combed the Internet and found five practical definitions from reputable resources:

"the most basic machine learning is the use of algorithms to parse data, learn from it, and then make decisions or predictions about something in the world." -Nvidia

"Machine learning is the science that allows computers to take action without explicit programming." -Stanford

"Machine learning is based on algorithms that can be learned from data without relying on rule-based programming." -McKinsey & Co.

"Machine learning algorithms can be used to figure out how to perform important tasks by promoting examples." -University of Washington

"the field of machine learning aims to answer the question:" how can we build computer systems that can automatically improve based on experience, and what is the basic law for managing all learning processes? "- Carnegie Mellon University

Basic concepts of machine learning

There are many different types of machine learning algorithms, hundreds of which are published every day, and they are usually based on learning style (i.e. supervised learning, unsupervised learning, semi-supervised learning) or through formal or functional similarity (i.e. classification, regression, decision tree, clustering, deep learning, etc.). Regardless of learning style or function, all combinations of machine learning algorithms include the following:

To represent (a group of classifiers or computer-understood languages).

Evaluation (also known as objective / scoring function)

Optimization (search method; usually the classifier with the highest score; using off-the-shelf and custom optimization methods)

Visual representation of Machine Learning Model

So far, concepts and points can only be used for understanding. When people ask, "what is machine learning?" They often want to see what it is and what it does. Here are some visual representations of the machine learning model, along with links for more information. More resources can be found at the bottom of this article.

Decision tree model

Gaussian mixture model

Dropout neural network

Combining chromaticity and luminance using convolution neural network

How do we make machines learn?

There are many different ways to make machine learning, from using a basic decision tree to clustering to the artificial neural network layer (which has given way to deep learning), depending on the task and type you want to accomplish and the amount of data available to you. This dynamic plays a role in a variety of applications, such as medical diagnosis or self-driving cars.

Although the focus is usually on choosing the best learning algorithm, the researchers found that some of the most interesting problems stem from available machine learning algorithms. In most cases, this is a problem with training data, but this also happens when machine learning is used in new domains.

Research done when dealing with real applications usually drives progress in this area for two reasons: 1. Trends in discovering the boundaries and limitations of existing methods 2. Researchers and developers work with domain experts to use time and expertise to improve system performance.

Sometimes this is also caused by an accident. We may consider a set of models or a combination of many learning algorithms to improve accuracy, which is an example. The 2009 Netflix Price team found that when they combined learners with learners from other teams, they achieved the best results, thus improving the recommendation algorithm.

In terms of business and other applications, an important point of view (based on interviews and conversations with experts in this field) is that machine learning is not only, even automated, which is a concept that is often misunderstood. If you think so, you are bound to miss out on the valuable insights and opportunities that machines can provide (rethink the entire business model, such as manufacturing and agriculture).

Learning machines are useful to humans because they have all the processing power and can more quickly highlight or find patterns in large (or other) data that humans may miss. Machine learning is a tool that can be used to enhance human ability to solve problems and make wise inferences from a wide range of problems, from helping to diagnose diseases to proposing solutions to global climate change.

Challenges and limits

"Machine learning cannot be built out of nothing. Its function is to get more from less." -Dr. Pedro Domingo, University of Washington.

The two biggest, historic (and persistent) problems in machine learning involve overfitting (in which the model shows a bias against training data and does not extend to new data, and / or changes. That is, learning random things when training new data) and dimensions (algorithms with more features work on higher / more dimensions, making it more difficult to understand the data). In some cases, accessing a sufficiently large dataset is also a major problem.

One of the most common mistakes among machine learning beginners is to successfully test training data with the illusion of success; Domingo (and others) emphasizes the importance of keeping some data sets separate when testing models, and only use that retained data to test the selected model, and then learn the entire data set.

When learning algorithms (that is, learners) do not work, the usually faster way to success is to provide more data to machines, and its availability has become the main driver of recent advances in machine and deep learning algorithms. Year; however, this can lead to scalability problems, we have more data, but having time to understand the data is still a problem.

Deep Learning and Modern Development of Neural Network

Deep learning involves the research and design of machine algorithms for learning good representations of data at multiple levels of abstraction (the way computer systems are arranged). The recent promotion of deep learning through DeepMind, Facebook and other institutions highlights it as the "next frontier" of machine learning.

The International Conference on Machine Learning (ICML) is widely regarded as one of the most important opportunities in the world. They are committed to solving the current challenges of deep learning:

Unsupervised Learning in small data sets

Simulation-based learning and transferability to the real world

In the past decade, deep learning system has made great progress in such fields as object detection and recognition, text to speech, information retrieval and so on. Research is now focused on developing data-efficient machine learning, that is, deep learning systems, which can learn more efficiently in less time and less data, in cutting-edge areas such as personalized health care, robot reinforcement learning, emotional analysis, and so on.

This is the end of the article on "what are the visual representations of machine learning?" Thank you for reading! I believe you all have a certain understanding of the knowledge of "what is the visual representation of machine learning". If you want to learn more knowledge, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.