Xxii республикалық студенттер мен жас ғалымдардың ғылыми конференция материалдары
жүктеу/скачать 8,13 Mb. Pdf көрінісі
|
Сборник материалов конференции (продолжение)
| Support vector machines
The Support Vector Machines Method or SVM is a linear algorithm used in classification and regression problems. This algorithm is widely used in practice and can solve both linear and nonlinear problems. The essence of the “Machines” of Support Vectors is simple: the algorithm creates a line or hyperplane that divides the data into classes. The main task of the algorithm is to find the most correct line, or hyperplane, dividing the data into two classes. SVM is an algorithm that receives data at the input and returns such a dividing line. The SVM algorithm is arranged in such a manner that it looks for points on the graph that are located straight to the separation line in the closest way. These points are called support vectors. Then, the algorithm calculates the distance between the support vectors and the dividing plane. This is the distance called the gap. The major intention of the algorithm is to maximize the clearance distance. The best hyperplane is considered to be a hyperplane for which this gap is as 230 large as possible (Figure 5). Concrete and applied use case of SVM is solving the problem of classification for ultrasonic medicine image[5]. Figure 5 Conclusion Sad reality is that we can meet practically applied statistical learning models in very rare cases. Although using statistical learning techniques already become the integral part of other spheres, such as banking system, insurance, internet recommendation system, marketing and etc. But there is no the same success in the field of medicine. Why it happens? First of all medicine datasets are very specific and usually confidential. So researchers have not too much open accessible datasets for increasing their competence level of building effective models for medicine. At second, medical datasets commonly not stand out for good quality, there are usually many missed values, incorrect values and insufficiency of informative inputs. And finally, the last one, but not by the importance, it is vital requirement in high competencies in two different spheres – medical and data science. Any predicted output have to be justified and checked out not only from mathematical-statistical perspective, but also from biological-medical aspect. Hope this article will attract and inspire people for investigating this crucial topic and find new answers of effective implementing statistical learning models in the healthcare. жүктеу/скачать 8,13 Mb. Достарыңызбен бөлісу:
|