Abstract: Systems and methods for training models to improve fairness.

Abstract: A computing device receives, via a communication network, map data including (i) a description of geometries of map features and (ii) a first description of visual characteristics defined separately and independently of the description of the geometries. The computing device applies the visual characteristics to the geometries to render a first digital map. The computing device then receives, via the communication network, a second description of visual characteristics for application to the geometries previously provided to the computing device as part of the map data, and applies the second visual characteristics to the previously received geometries of the plurality of map features to render a second digital map.

Abstract: To organize configuration parameters for rendering map features on digital maps, a data structure storing multiple nodes is generated. Respective sets of map feature attributes and configuration parameters are assigned to the nodes. The configuration parameters specify visual attributes for rendering map features. Indications of relationships between the nodes are generated, where at least several of the nodes include several child nodes. An indication of one or more map feature attributes to which a certain map feature belongs is received, and the data structure is traversed according to the indicated relationships between the nodes to generate a set of configuration parameters for rendering the map feature. Traversing the data structure includes comparing the indicated map feature attribute or map feature attributes to sets of map feature attributes at some of the nodes, such that the generated set of configuration parameters includes configuration parameters from at least two of the nodes.

Abstract: First map data is generated in a vector format for displaying a first map image at a client device. The first map image corresponds to a perspective of a camera panned to a certain location at a certain zoom level. The first map data is provided to the client device via a communication network. Modification data is generated at a map server in a vector format, the modification data specifying a difference between the first map data and a second map data, where the second map data is for displaying a second map image corresponding to the same perspective of the camera panned to the same location at the same zoom level. The modification data is provided to the client device via the communication network for generating the second map image using the first map data and the modification data.

Abstract: To generate a description of a raster map image that includes a representation of a body of water having a color gradient that appears constant in response to a zoom operation, a description of geometry of the body of water is generated for a selected geographic area. Using a raster representation of the body of water having a color gradient, at least two blur raster images are generated: a first blur raster image is generated using a first blur radius, and a second blur raster image is generated using a second blur radius different from the first blur radius. The description of the geometry of the body of water, the first blur raster image, and the second blur raster image are provided to a client device for generating raster map images of the selected geographic area.

Abstract: First map data is generated in a vector format for displaying a first map image at a client device. The first map image corresponds to a perspective of a camera panned to a certain location at a certain zoom level. The first map data is provided to the client device via a communication network. Modification data is generated at a map server in a vector format, the modification data specifying a difference between the first map data and a second map data, where the second map data is for displaying a second map image corresponding to the same perspective of the camera panned to the same location at the same zoom level. The modification data is provided to the client device via the communication network for generating the second map image using the first map data and the modification data.

Abstract: A map server generates vector descriptors in a non-raster format, each indicating a geometry of a respective map element, to render a first map image of a selected geographic region at a client device. The map server then provides the vector descriptors to the client device. Upon receiving an indication that a second map image for the selected geographic region is to be rendered at the client device, the map server generates a modification indication that indicates a set of one or more vector descriptors, each of which corresponds to a respective map element that is rendered in one but not both of the first map image and the second map image, and provides the modification indication to the client device, so that the client device can render the second map image using at least some of the vector descriptors and the modification indication.

Abstract: To generate a description of a raster map image that includes a representation of a body of water having a color gradient that appears constant in response to a zoom operation, a description of geometry of the body of water is generated for a selected geographic area. Using a raster representation of the body of water having a color gradient, at least two blur raster images are generated: a first blur raster image is generated using a first blur radius, and a second blur raster image is generated using a second blur radius different from the first blur radius. The description of the geometry of the body of water, the first blur raster image, and the second blur raster image are provided to a client device for generating raster map images of the selected geographic area.

Abstract: A computing device receives, via a communication network, map data including (i) a description of geometries of map features and (ii) a first description of visual characteristics defined separately and independently of the description of the geometries. The computing device applies the visual characteristics to the geometries to render a first digital map. The computing device then receives, via the communication network, a second description of visual characteristics for application to the geometries previously provided to the computing device as part of the map data, and applies the second visual characteristics to the previously received geometries of the plurality of map features to render a second digital map.

Abstract: A map server generates vector descriptors and corresponding style identifiers for rendering a map image. Each vector descriptor indicates a geometry of a respective map element in accordance with a vector graphics format, and each style identifier identifies a style according to which the corresponding map element is rendered. The map server then causes map data including the vector descriptors and the style identifiers to be transmitted to the client device via a communication network. When the map server receives a request for style data for modifying a map image at the client device, the map server generates a description of visual characteristics for one or more styles identified by the style identifiers included in the map data previously provided to the client device and causes the description of visual characteristics to be transmitted to the client device for applying to the vector descriptors at the client device.

Abstract: A digital image rendering system, such as a geographic map rendering system, receives image data from an image database, such as a map database, in the form of data having image features defined as sets of image objects arranged in a series of layers. The image rendering system processes the received data in a manner that allows for rendering the original layered data without rendering each original layer separately by reordering the original layered digital image data into fewer layers.

Abstract: A digital image rendering system, such as a geographic map rendering system, receives image data from an image database, such as a map database, in the form of data having image features defined as sets of image objects arranged in a series of layers. The image rendering system processes the received data in a manner that allows for rendering the original layered data without rendering each original layer separately by reordering the original layered digital image data into fewer layers.

Abstract: A map server generates vector descriptors in a non-raster format, each indicating a geometry of a respective map element, to render a first map image of a selected geographic region at a client device. The map server then provides the vector descriptors to the client device. Upon receiving an indication that a second map image for the selected geographic region is to be rendered at the client device, the map server generates a modification indication that indicates a set of one or more vector descriptors, each of which corresponds to a respective map element that is rendered in one but not both of the first map image and the second map image, and provides the modification indication to the client device, so that the client device can render the second map image using at least some of the vector descriptors and the modification indication.