In the study of learning strategy, the definition of learning strategies has always been a fundamental problem. What are the strategies for learning, people from different research angles and using different research methods, put forward their different views. Learning Machine Learning is a complex intelligence activities, learning and reasoning are closely linked, in order to improve the machine learning system, we have many ways to improve the machine learning system, such as adding more training set, modify neural network the structure and so on. So how learning strategies based on machine learning to classify? Here we give you answer these questions.
In machine learning, learning strategies can not be ignored is the content, learning strategy refers to the process of reasoning strategies learning system used. And generally speaking a learning system always consists of two parts and learning environment. Information provided by the environment, the learning part is achieved conversion information, using understandable form memorized, and obtain useful information. This knowledge is learning strategies. In general, the classification is based on students' learning strategies to realize how much information is required for the conversion and ease of reasoning to the classification, compliance simple to complex, divided into the following six basic types from less to more order.
The first is to teach, students to obtain information from the environment, to convert the knowledge gained can be used to form the interior of the new knowledge and existing knowledge organically combined into one. Teachers propose and organize knowledge in some form, so that students have the knowledge may continue to increase. It requires students to have a certain degree of reasoning ability, but the environment is still a lot of work. Similar schools teaching this way of learning and of human society, the learning task is to build a system to enable it to accept the teachings and recommendations, and effective storage and application of knowledge learned. Many expert systems use this approach in establishing the knowledge base to achieve knowledge acquisition.
The second is rote learning, in which learners learn without any reasoning or other knowledge transfer, directly absorb the information provided by the environment. Such learning system the main consideration is how to index stored knowledge and use it. Learning system is pre-programmed by direct learning is good, good program structure, learners do not make any work, or learning by directly receiving the established facts and data, input information without any reasoning.
The third is learning by analogy, the analogy is an important method for people to understand the world, is also an important means of inducing people to learn new things, be creative thinking. By analogy, the knowledge derived from the corresponding source domain knowledge of the target domain, in order to achieve learning. Analogy learning system can make an existing computer applications to adapt into new areas, to complete a similar function not originally designed.
The fourth is the interpretation of the study, it refers to the process of learning, reasoning used by students as a form of deductive reasoning. Reasoning from axioms to observe and ask questions based on the analysis, the conclusions deduced through logical transformation. This reasoning is fidelity and specialized conversion process, so that students can get useful knowledge in the reasoning process to master the theoretical knowledge and skills. This learning method comprising macro operations study, editing and block technical knowledge. The reverse process of deductive reasoning is inductive reasoning.
The fifth is the explanation-based learning, students according to teachers of the target concept, an example of this concept, field theory and operational guidelines, first construct an explanation as to why this example to meet the target concept and then explain the concept of targeted promotion a sufficient condition to meet the operational criteria. EBL has been widely used in the knowledge base refinement and improve system performance.
The sixth is the inductive learning, inductive learning aims to generalize extracted general judging rules and patterns from a large number of empirical data, are derived from the special case of a general rule of learning. This reasoning effort to learn far more than teaching and learning deductive learning, because the environment does not provide a general description of the concept. To some extent, the amount of inductive reasoning is also larger than the analogy of learning to learn, because there is a similar concept can be drawn as a source concept. Inductive learning is the most basic, but also the development of a more mature approach to learning in the field of artificial intelligence has been widely studied and applied.
By further understanding of machine learning, machine learning we have a more step to deepen understanding. By article describes, we also understand how learning strategies based on machine learning to classify. These questions may seem difficult, but knowledge of machine learning are little by little, we are required depth slowly. I hope everyone can learn something layers, glory to play in the workplace.