Scientific and technological developments appear intelligent bus card greatly facilitates our daily lives, Metro bus can be used, instead of the wallet can even make a payment, these features we have come to realize over daily, but the smart bus cards can catch the thief, you heard ? However, it is really realized in real life.
This smart bus card does not shine brightly on your eyes
Prof. Xiong Hui of Rutgers University and others reported the results at a recent knowledge discovery and data mining conference in San Francisco. The principle is as follows: The vast majority of passengers will choose the best travel plan when they are traveling by bus or subway. The travel time is the shortest, or the minimum number of transfers; but there are very few people who ride on the route is very strange, for example, will bypass one There is no regularity in the big circle or the sudden change in the route of travel. If someone has a lot of abnormal behavior, he may be a thief.
The truth may seem simple, but it is difficult to find a real thief. Researchers point out that the automatic ticketing system collects the massive travel records of millions of passengers. Only a very few of them may be thieves. Identifying such a small group of people in such large-scale data is no different from a needle in a haystack.
To this end, the researchers analyzed the data records of approximately 1.6 billion bus card swipes from April to June 2014 in Beijing in two steps, involving approximately 6 million passengers. In the first step, they divided Beijing into small functional blocks such as housing, work, education, shopping, entertainment, and medical care, and established 896 bus routes, nearly 45,000 bus stops and 18 subway lines, and 320 buses. The subway station's public transportation network dataset filters out ordinary passengers from the huge bus card records through data modeling; the second step combines the theft information collected from police reports and Weibo, through machine learning algorithms from abnormalities. The potential thieves were discovered in the travel information.
The results show that if the thief is later confirmed to be verified, the accuracy of the "abnormal behavior" according to the above method can be as high as 92.7%. However, the accuracy of the reverse is somewhat lower: Only 14 suspicious individuals with “behavioural anomalies†were screened out, and only 1 was subsequently identified as a thief.
Despite this, Xiong Hui believes that using closed-circuit cameras to monitor a small number of suspicious individuals is far more efficient than tracking millions of passengers. But what if the thief frequently uses public transportation cards? He said that even if there are ways to change cards, such as thieves often gang activities, this is also an obvious feature.
Some experts questioned this technology. The British "The Economist" magazine cited the Chief Technology Officer of the London Transport Authority Shashi Weierma, saying that from the relevant records in London, a large number of ordinary passengers will also have a variety of "weird, exciting and complex" behavior when they travel. It is not as easy as it sounds to filter out a handful of criminals out of a large number of passengers.
This smart bus card does not shine brightly on your eyes
Prof. Xiong Hui of Rutgers University and others reported the results at a recent knowledge discovery and data mining conference in San Francisco. The principle is as follows: The vast majority of passengers will choose the best travel plan when they are traveling by bus or subway. The travel time is the shortest, or the minimum number of transfers; but there are very few people who ride on the route is very strange, for example, will bypass one There is no regularity in the big circle or the sudden change in the route of travel. If someone has a lot of abnormal behavior, he may be a thief.
The truth may seem simple, but it is difficult to find a real thief. Researchers point out that the automatic ticketing system collects the massive travel records of millions of passengers. Only a very few of them may be thieves. Identifying such a small group of people in such large-scale data is no different from a needle in a haystack.
To this end, the researchers analyzed the data records of approximately 1.6 billion bus card swipes from April to June 2014 in Beijing in two steps, involving approximately 6 million passengers. In the first step, they divided Beijing into small functional blocks such as housing, work, education, shopping, entertainment, and medical care, and established 896 bus routes, nearly 45,000 bus stops and 18 subway lines, and 320 buses. The subway station's public transportation network dataset filters out ordinary passengers from the huge bus card records through data modeling; the second step combines the theft information collected from police reports and Weibo, through machine learning algorithms from abnormalities. The potential thieves were discovered in the travel information.
The results show that if the thief is later confirmed to be verified, the accuracy of the "abnormal behavior" according to the above method can be as high as 92.7%. However, the accuracy of the reverse is somewhat lower: Only 14 suspicious individuals with “behavioural anomalies†were screened out, and only 1 was subsequently identified as a thief.
Despite this, Xiong Hui believes that using closed-circuit cameras to monitor a small number of suspicious individuals is far more efficient than tracking millions of passengers. But what if the thief frequently uses public transportation cards? He said that even if there are ways to change cards, such as thieves often gang activities, this is also an obvious feature.
Some experts questioned this technology. The British "The Economist" magazine cited the Chief Technology Officer of the London Transport Authority Shashi Weierma, saying that from the relevant records in London, a large number of ordinary passengers will also have a variety of "weird, exciting and complex" behavior when they travel. It is not as easy as it sounds to filter out a handful of criminals out of a large number of passengers.
Matting Agent For Electrophoretic Coatings
Matting Agent For Electrophoretic Coatings,E Coating Matting Agent,Electrophoretic Coatings Silica,White Powder Silica Matting Agent
Guangzhou Quanxu Technology Co Ltd , https://www.skychema.com