Techniques are disclosed for generating a sequence of symbols based on input data for a neuro-linguistic model. The model may be used by a behavior recognition system to analyze the input data. A mapper component of a neuro-linguistic module in the behavior recognition system receives one or more normalized vectors generated from the input data. The mapper component generates one or more clusters based on a statistical distribution of the normalized vectors. The mapper component evaluates statistics and identifies statistically relevant clusters. The mapper component assigns a distinct symbol to each of the identified clusters.

Systems, methods, tangible non-transitory computer-readable media, and devices associated with object localization and generation of compressed feature representations are provided. For example, a computing system can access training data including a source feature representation and a target feature representation. An encoded target feature representation can be generated based on the target feature representation and a machine-learned encoding model. A binarized target feature representation can be generated based on the encoded target feature representation and lossless binarization operations. A reconstructed target feature representation can be generated based on the binarized target feature representation and a machine-learned decoding model. A matching score for the source feature representation and the reconstructed target feature representation can be determined. A loss associated with the matching score can be determined. Parameters of the machine-learned encoding model and the machine-learned decoding model can be adjusted based on the loss.

Systems, methods, tangible non-transitory computer-readable media, and devices associated with object localization and generation of compressed feature representations are provided. For example, a computing system can access source data and target data. The source data can include a source representation of an environment including a source object. The target data can include a compressed target feature representation of the environment. The compressed target feature representation can be based on compression of a target feature representation of the environment produced by machine-learned models. A source feature representation can be generated based on the source representation and the machine-learned models. The machine-learned models can include machine-learned feature extraction models or machine-learned attention models. A localized state of the source object with respect to the environment can be determined based on the source feature representation and the compressed target feature representation.

A client device configured with a neural network includes a processor, a memory, a user interface, a communications interface, a power supply and an input device, wherein the memory includes a trained neural network received from a server system that has trained and configured the neural network for the client device. A server system and a method of training a neural network are disclosed.

Provided are a method for training a font generation model, a method for establishing a font library, and a device. The method for training a font generation model includes the following steps. A source-domain sample character is input into the font generation model to obtain a first target-domain generated character. The first target-domain generated character is input into a font recognition model to obtain the target adversarial loss of the font generation model. The model parameter of the font generation model is updated according to the target adversarial loss.

There is provided an image-based information extraction model, method, and apparatus, a device, and a storage medium, which relates to the field of artificial intelligence (AI) technologies, specifically to fields of deep learning, image processing, computer vision technologies, and is applicable to optical character recognition (OCR) and other scenarios. A specific implementation solution involves: acquiring a to-be-extracted first image and a category of to-be-extracted information; and inputting the first image and the category into a pre-trained information extraction model to perform information extraction on the first image to obtain text information corresponding to the category.

In some examples, dark web content analysis and identification may include ascertaining data that includes text and images, and analyzing the data by performing deep learning based text and image processing to extract text embedded in the images, and deep embedded clustering to generate clusters. Clusters that are to be monitored may be ascertained from the generated clusters. A determination may be made as to whether the ascertained data is sufficient for classification. If so, a deep convolutional generative adversarial networks (DCGAN) based detector may be utilized to analyze further data with respect to the ascertained clusters, and alternatively, a convolutional neural network (CNN) based detector may be utilized to analyze the further data with respect to the ascertained clusters. Based on the analysis of the further data, an operation associated with a website related to the further data may be controlled.

The goal of this invention is to develop smart and fast data processing scheme for more computational efficient deep learning to support adaptive and real-time applications. We propose to apply Singular-Value Decomposition (SVD)-QR algorithm to preprocessing of deep learning for large scale data input. For the mass data input, we apply Limited Memory Subspace Optimization for SVD (LMSVD)-QR algorithm to increase the data processing speed. Simulation results in automated handwritten digit recognition show that SVD-QR and LMSVD-QR can tremendously reduce the number of input to deep learning neural network without losing its performance, and both can tremendously increase the data processing speed for deep learning.

A handwriting recognition method which uses an invertible label embedding (encoding) algorithm to embed character strings into an Euclidean vector space as attribute vectors, uses a CNN to learn and predict attribute vectors of handwriting images in this Euclidean vector space, and then directly decodes a predicted attribute vector into a character string using a decoding algorithm that is the inverse of the invertible encoding algorithm. No lexicon is required to decode the predicted attribute vector. Thus, this method can recognize images containing handwritten digital sequences commonly encountered in many practical applications, such as quantities, dollar, date, phone number, social security numbers, zip code, etc. which are outside of common lexicons.

A system and method for field extraction including determining a key position of a key in an electronic file, isolating candidate key values based on a distance from the key position, selecting a key value from the candidate key values based on an output of a trained neural network, and extracting the key and the key value from the electronic file, regardless of a key-value structure.