Traffic condition forecasting using matrix compression and deep neural networks
Traffic condition forecasting techniques are provided that use matrix compression and deep neural networks. An illustrative method comprises obtaining a compressed origination-destination matrix indicating a cost to travel between pairs of a plurality of nodes, wherein the compressed origination-destination matrix is compressed using a locality-aware compression technique that maintains only non-empty data; obtaining a trained deep neural network trained using the compressed origination-destination matrix and past observations of traffic conditions at various times; and applying traffic conditions between two nodes in the compressed origination-destination matrix at a time, t, to the trained deep neural network to obtain predicted traffic conditions between the two nodes at a time, t+Δ. A tensor can be generated indicating an evolution of traffic conditions over a time span using a stacked Origination-Destination matrix comprising a plurality of past observations representing the time span.
1. A method, comprising:
obtaining a compressed origination-destination matrix indicating a cost to travel between pairs of a plurality of nodes, wherein said compressed origination-destination matrix is compressed using a locality-aware compression technique that maintains only non-empty data, wherein the compressed origination-destination matrix is created by:
selecting a point-of-interest;
creating a distance vector only for said non-empty data; and
applying a space-filling curve technique from the selected point-of-interest to said distance vector;
obtaining a trained deep neural network trained using said compressed origination-destination matrix and past observations of traffic conditions at various times; and
applying traffic conditions between two nodes in said compressed origination-destination matrix at a time, t, to said trained deep neural network to obtain predicted traffic conditions between said two nodes at a time, t+Δ.
2. The method of claim 1 , wherein said predicted traffic conditions comprise one or more of a travel time and a cost-in-time between any pair of nodes.
3. The method of claim 1 , wherein said compressed Origination-Destination matrix comprises a stacked origination-destination matrix comprising a plurality of past observations.
4. The method of claim 3 , further comprising generating a tensor using said stacked origination-destination matrix indicating an evolution of traffic conditions over a time span represented by said plurality of past observations.
5. The method of claim 1 , wherein said traffic conditions at a time, t, are compressed using said locality-aware compression technique.
6. The method of claim 1 , wherein said training of said deep neural network comprises a supervised learning using said compressed origination-destination matrix with additional compressed origination-destination matrices representing said past observations of traffic conditions at various times.
7. The method of claim 1 , further comprising predicting a number of predefined traffic disruptions in a given area in a predefined time period based on a predicted number of vehicles in said given area obtained using said trained deep neural network.
8. The method of claim 1 , further comprising predicting a traffic condition in a given area in a predefined time period based on a predicted number of vehicles in said given area obtained using said trained deep neural network.
9. The method of claim 1 , further comprising identifying a top N list of paths between two nodes that are most impacted by one or more predefined traffic disturbances using said trained deep neural network.
10. The method of claim 1 , further comprising forecasting one or more geographic areas having one or more predefined traffic disturbances in a near future using said trained deep neural network.
11. A computer program product, comprising a non-transitory machine-readable storage medium having encoded therein executable code of one or more software programs, wherein the one or more software programs when executed by at least one processing device perform the following steps:
obtaining a compressed origination-destination matrix indicating a cost to travel between pairs of a plurality of nodes, wherein said compressed origination-destination matrix is compressed using a locality-aware compression technique that maintains only non-empty data, wherein the compressed origination-destination matrix is created by:
selecting a point-of-interest;
creating a distance vector only for said non-empty data; and
applying a space-filling curve technique from the selected point-of-interest to said distance vector;
obtaining a trained deep neural network trained using said compressed origination-destination matrix and past observations of traffic conditions at various times; and
applying traffic conditions between two nodes in said compressed origination-destination matrix at a time, t, to said trained deep neural network to obtain predicted traffic conditions between said two nodes at a time, t+Δ.
12. An apparatus, comprising:
a memory; and
at least one processing device, coupled to the memory, operative to implement the following steps:
obtaining a compressed origination-destination matrix indicating a cost to travel between pairs of a plurality of nodes, wherein said compressed origination-destination matrix is compressed using a locality-aware compression technique that maintains only non-empty data, wherein the compressed origination-destination matrix is created by:
selecting a point-of-interest;
creating a distance vector only for said non-empty data; and
applying a space-filling curve technique from the selected point-of-interest to said distance vector;
obtaining a trained deep neural network trained using said compressed origination-destination matrix and past observations of traffic conditions at various times; and
applying traffic conditions between two nodes in said compressed origination-destination matrix at a time, t, to said trained deep neural network to obtain predicted traffic conditions between said two nodes at a time, t+Δ.
13. The apparatus of claim 12 , wherein said compressed origination-destination matrix comprises a stacked origination-destination matrix comprising a plurality of past observations.
14. The apparatus of claim 13 , further comprising generating a tensor using said stacked origination-destination matrix indicating an evolution of traffic conditions over a time span represented by said plurality of past observations.
15. The apparatus of claim 12 , wherein said training of said deep neural network comprises a supervised learning using said compressed origination-destination matrix with additional compressed origination-destination matrices representing said past observations of traffic conditions at various times.
16. The apparatus of claim 12 , further comprising predicting one or more of a number of predefined traffic disruptions and a traffic condition in a given area in a predefined time period based on a predicted number of vehicles in said given area obtained using said trained deep neural network.
17. The apparatus of claim 12 , further comprising identifying a top N list of paths between two nodes that are most impacted by one or more predefined traffic disturbances using said trained deep neural network.
18. The apparatus of claim 12 , further comprising forecasting one or more geographic areas having one or more predefined traffic disturbances in a near future using said trained deep neural network.
19. The computer program product of claim 11 , further comprising predicting a number of predefined traffic disruptions in a given area in a predefined time period based on a predicted number of vehicles in said given area obtained using said trained deep neural network.
20. The computer program product of claim 11 , further comprising predicting a traffic condition in a given area in a predefined time period based on a predicted number of vehicles in said given area obtained using said trained deep neural network.