Lei Feng Network (search "Lei Feng Network" public concern) : Ye Jieping, male, Ph.D., University of Minnesota, the current Dick Research Institute vice president. He is a tenured professor at the University of Michigan and a member of the Management Committee at the University of Michigan Big Data Research Center. Ye Jieping is an international leader in machine learning. He is mainly engaged in the fields of machine learning, data mining, and big data analysis. He is a world leader in large-scale sparse model learning.
Ye Jiping, vice president of Dripping Research Institute, lectures at KDD
On August 17th in Beijing, the international top data mining conference KDD 2016 was opened in San Francisco, USA. Technology companies including Google, Facebook, Microsoft, Amazon, Alibaba, Tencent, Baidu and Didi traveled to the conference. Ye Jieping, vice president of Dripping Research Institute, delivered a speech at the conference on how to use big data for smart scheduling and supply and demand forecasting.
Problems and ChallengesYe Jieping: “The biggest asset of DDT is a large amount of travel big data generated every day. We handle more than 70TB of data per day, 9 billion route planning requests, 9 billion map positioning, 9 billion map positioning, and 1 billion dispatches. This is still We have acquired data from Uber China before. For us, how to use such a large amount of traffic data is a major challenge."
Challenge 1: Long wait time - huge order volume
Ye Jieping: “Didi Chuxing was founded in 2012 and currently provides taxis, special trains, express trains, and rides. We completed 1.43 billion travel orders last year, which is almost twice the amount of all taxi orders in the US in 2015. 2016 In March, we achieved a new milestone, with more than 10 million orders for sunrise orders, which is equivalent to 5 to 6 times the total number of trip orders shared by the United States on a single day.â€
Challenge 2: Traffic jams - high vehicle density
Ye Jieping: “Traffic congestion is a big problem, and China is no exception. The ten cities with the highest density of vehicles in the world, the first eight from China, followed by Tokyo and New York.â€
Challenge 3: Difficulties in taxiing - imbalance between supply and demand
Ye Jieping: "People may find it difficult to hit the car during peak hours. This is caused by the imbalance between supply and demand."
Artificial intelligence-based solutionsYe Jieping: "We use the machine learning model to look for patterns from the mass of travel data. The most important thing is to find the most effective machine learning model and feature mining to solve ETA problems. We just spent a lot of time looking for features when we first started modeling for ETA. , Now we are ready to continuously optimize our model, will continue to improve the accuracy of the estimates, better service users."
Didi Chuxing is the first company in China to successfully apply machine learning to ETA. This is the key technology to solve the problem of "order efficient matching" and "driver dispatch."
Intelligent dispatch
Intelligent dispatch is one of the core technologies for Didi operations. Ye Jieping said: “Every second, we have to match thousands of passengers and drivers. The distance between passengers and drivers or driving time is a measure of the quality of the dispatch. Main Indicators We need to use two key map technologies, namely route planning and ETA (estimate the travel time required for any end point) to complete the dispatch.
How to implement smart dispatch
The traditional method generally calculates the time through the road conditions and the average speed of each section of the road, and then adds the possible waiting time to obtain the overall required time, while the drop is to use machine learning to calculate the time, which greatly improves the user experience. According to this technology, the current distance and the time to reach the end point can be updated in real time on the dribble trip platform.
Supply and demand forecast
For the problem of imbalance between supply and demand, Ye Jieping said: "A better solution may be to forecast the supply and demand situation so as to intelligently dispatch drivers in advance. For example, we predict that a certain region will have a large imbalance between supply and demand. We will Drivers will be sent to this area to avoid the inability of users to meet passenger demand. Realizing the supply and demand forecast will bring three major benefits: balance between supply and demand, improved passenger car experience, and increased driver income."
Intelligent scheduling
It is logical to create portraits for each driver, including their habits of receiving orders and receiving orders, matching the right orders to the driver's hands, ensuring that both drivers and users find the service they like.
How to implement intelligent scheduling
At the KDD meeting, Ye Jieping also disclosed that the company is developing a visual system named "Nine Dragons," Duse-eye. "The system can show what has happened in the past and what is happening, such as telling us where there is traffic congestion and the current Supply and demand, etc."
Smart Carpooling Program
Another solution to the difficulty of getting cars is carpooling. Yepping said that he can intelligently optimize the carpooling scheme through machine learning. He mentioned: “Carpooling reduces people’s travel costs and car fuel costs, but the key issue is that all passengers need to be The time spent is minimal, and it is clear that the more similar the route between passengers, the less time is needed.In addition, the question of how carpool pricing is also a problem, the key is to calculate the expected profit of each single, if the expected profit is high We will give higher discounts. This is actually a matter of machine learning."
Dripping ambition The Drip Research Institute is a brand-new and innovative research institution for DDT, and it is also the “brain†of the Drip Trip. In the future, all the technological innovations that help improve the efficiency of mobile travel will be hatched here.
At present, the research direction of Didi Research Institute includes: machine learning, computer vision, artificial intelligence, data mining, optimization theory, distributed computing, and so on. The Drip Research Institute is closely integrated with the business line, and each research result can be applied to the corresponding products at the fastest speed, and bring convenience to millions of users.
Merging Uber China, the network is about legal, and after it solves market competition and operational issues, Didi is ready to continue to lead the field with technology breakthroughs. Didi Research Institute is the product of its active deployment of artificial intelligence. In the coming DT era, there will be plenty of opportunities to catch up with the drops of artificial intelligence.
Lcd Tonch Screen For Iphone 6,Lcd Display For Iphone 6Sp,Mobile Lcd For Iphone 6Sp,Lcd Touch Screen For Iphone 6S
Shenzhen Xiangying touch photoelectric co., ltd. , https://www.starstpmobile.com