Abstract: A computer-implemented method of trajectory prediction includes obtaining a first cross-attention between a vectorized representation of a road map near a vehicle and information obtained from a rasterized representation of an environment near the vehicle by processing through a first cross-attention stage; obtaining a second cross-attention between a vectorized representation of a vehicle history and information obtained from the rasterized representation by processing through a second cross-attention stage; operating a scene encoder on the first cross-attention and the second cross-attention; operating a trajectory decoder on an output of the scene encoder; obtaining one or more trajectory predictions by performing one or more queries on the trajectory decoder.

Abstract: A method of predicting vehicle trajectory includes operating a scene encoder on an environmental representation surrounding a vehicle; concatenating an output of the scene encoder with a history trajectory; applying a sequence encoder to a result of the concatenating; refining an output of the sequence encoder based on the history trajectory; and generating one or more predicted future trajectories by operating a decoder on an output of the refining.

Abstract: A system for predicting aggressive driving behavior for a driver of a vehicle includes a first edge computing device that can acquire spatial-temporal data for the vehicle from one or more sensors that are part of traffic infrastructure. The first edge computing device includes a processor and instructions executable by the processor that execute deep learning methods on the data from the sensors to cluster the data as a driving score. A trained model is applied to the driving score to determine an aggressive driving behavior risk level, and the first edge computing device is configured to predict the aggressive driving behavior based on the aggressive driving behavior risk level.

Abstract: A pedestrian tracking system includes: a buffer or a memory configured to store a trajectory sequence of a pedestrian; a step attention module and a control module. The step attention module iteratively performs a step attention process to predict states of the pedestrian. Each iteration of the step attention process includes the step attention module: learning the stored trajectory sequence to provide time-dependent hidden states, reshaping each of the time-dependent hidden states to provide two-dimensional tensors; condensing the two-dimensional tensors via convolutional networks to provide convolutional sequences; capturing global information of the convolutional sequences to output a set of trajectory patterns represented by a new sequence of tensors; learning time-related patterns in the new sequence and decoding the new sequence to provide one or more of the states of the pedestrian; and modifying the stored trajectory sequence to include the predicted one or more of the states of the pedestrian.

Abstract: A system for predicting aggressive driving behavior for a driver of a vehicle includes a first edge computing device that can acquire spatial-temporal data for the vehicle from one or more sensors that are part of traffic infrastructure. The first edge computing device includes a processor and instructions executable by the processor that execute deep learning methods on the data from the sensors to cluster the data as a driving score. A trained model is applied to the driving score to determine an aggressive driving behavior risk level, and the first edge computing device is configured to predict the aggressive driving behavior based on the aggressive driving behavior risk level.

Abstract: Techniques for training a tokenizer (or word segmenter) are provided. In one technique, a tokenizer tokenizes a token string to identify individual tokens or words. A language model is generated based on the identified tokens or words. A vocabulary about an entity, such as a person or company, is identified. The vocabulary may be online data that refers to the entity, such as a news article or a profile page of a member of a social network. Some of the tokens in the vocabulary may be weighted higher than others. The language model accepts the weighted vocabulary as input and generates pseudo sentences. Alternatively, regular expressions are used to generate the pseudo sentences. The pseudo sentences are used to train the tokenizer.

Abstract: Techniques for training a tokenizer (or word segmenter) are provided. In one technique, a tokenizer tokenizes a token string to identify individual tokens or words. A language model is generated based on the identified tokens or words. A vocabulary about an entity, such as a person or company, is identified. The vocabulary may be online data that refers to the entity, such as a news article or a profile page of a member of a social network. Some of the tokens in the vocabulary may be weighted higher than others. The language model accepts the weighted vocabulary as input and generates pseudo sentences. Alternatively, regular expressions are used to generate the pseudo sentences. The pseudo sentences are used to train the tokenizer.

Abstract: Techniques for training a tokenizer (or word segmenter) are provided. In one technique, a tokenizer tokenizes a token string to identify individual tokens or words. A language model is generated based on the identified tokens or words. A vocabulary about an entity, such as a person or company, is identified. The vocabulary may be online data that refers to the entity, such as a news article or a profile page of a member of a social network. Some of the tokens in the vocabulary may be weighted higher than others. The language model accepts the weighted vocabulary as input and generates pseudo sentences. Alternatively, regular expressions are used to generate the pseudo sentences. The pseudo sentences are used to train the tokenizer.

Abstract: A system may include a network interface, a user interface, and a recommendation engine. The user interface may be configured to receive a job characteristic of a job profile of a job posted to the social network and a job bid from an entity related to job to the social network. The recommendation engine may be configured to determine an aggregate job score for the user based on a relevance of the job characteristic to a user characteristic and the job bid. The network interface may be configured to transmit a message related to the job to the user based, at least in part, on the aggregate job score.