30 results about "Self supervision" patented technology

Techniques for updating data operations in a perception system are discussed herein. A vehicle may use a perception system to capture data about an environment proximate to the vehicle. The perception system may output the data about the environment to a system configured to determine positions of objects relative to the perception system over time. The positions of the objects may be used to estimate an object velocity and may be compared against machine learning model outputs in a self-supervised manner to train the machine learning model to output object velocities based on inputs from the perception system. The output of the machine learning model may include a two-dimensional velocity for objects in the environment. The two-dimensional velocity may be used for a vehicle system such that the vehicle can make environmentally aware operational decisions, which may improve reaction time(s) and / or safety outcomes of the vehicle.

An example computer-implemented method for self-supervised training of an image processing model for histopathology images is provided. The example method includes obtaining a reference histopathology image; generating an augmented histopathology image, wherein generating the augmented histopathology image comprises performing, for an input image, at least one of the following augmentations: applying a blur to the input image and injecting noise artifacts into the blurred input image; or cropping a plurality of portions from the input image, wherein the plurality of portions are determined based on a minimum overlap criterion that has been updated over one or more iterations; and training the image processing model based on a similarity of latent representations generated by the image processing model respectively for the reference histopathology image and the augmented histopathology image.

An AI-based method for visual inspection of parts manufactured on a shop floor includes acquiring a set of real images of nominal parts manufactured on the shop floor to create training datasets. A self-supervised pre-trainer module is used to pre-train a loss computation neural network in a self-supervised learning process using a first dataset on pretexts defined by real-world conditions pertaining to the shop floor. The first dataset is labeled by automatically extracting pretext-related information from image metadata. A main anomaly trainer module is used to train a main anomaly detection neural network to reconstruct a nominal part image from an input manufactured part image in an unsupervised learning process using a second dataset. The main anomaly training measures a perceptual loss between an input image and a reconstructed image by measuring a difference between feature representations thereof at one or more layers of the pre-trained loss computation neural network.

Denoised magnetic resonance images are generated using a two-step process. An initial set of images is first denoised on a per channel basis using a locally low-rank-based denoising technique. The denoised coil channel images are transformed back into k-space and the denoised k-space data are then applied to a nonlinear image reconstruction. In some instances, the nonlinear image reconstruction can be implemented using a trained neural network. The neural network may be trained using a self-supervised learning technique.

Techniques are described for learning feature representations of medical images using a self-supervised learning paradigm and employing those feature representations for automating downstream tasks such as image retrieval, image classification and other medical image processing tasks. According to an embodiment, computer-implemented method comprises generating alternate view images for respective medical images included in set of training images using one or more image augmentation techniques or one or more image selection techniques tailored based on domain knowledge associated with the respective medical images. The method further comprises training a transformer network to learn reference feature representations for the respective medical images using their alternate view images and a self-supervised training process. The method further comprises storing the reference feature representations in an indexed data structure with information identifying the respective medical images that correspond to the reference feature representations.

A method for controlling a vehicle in an environment includes generating, via a cross-attention model, a cross-attention cost volume based on a current image of the environment and a previous image of the environment in a sequence of images. The method also includes generating combined features by combining cost volume features of the cross-attention cost volume with single-frame features associated with the current image. The single-frame features may be generated via a single-frame encoding model. The method further includes generating a depth estimate of the current image based on the combined features. The method still further includes controlling an action of the vehicle based on the depth estimate.

Provided is a process that includes obtaining data indicative of state of a dynamic mechanical system and an environment of the dynamic mechanical system, the data comprising a plurality of channels of data from a plurality of different sensors including a plurality of cameras and other sensors indicative of state of actuators of the dynamic mechanical system; forming a training set from the obtained data by segmenting the data by time and grouping segments from the different channels by time to form units of training data that span different channels among the plurality of channels; training a metric learning model to encode inputs corresponding to the plurality of channels as vectors in an embedding space with self-supervised learning based on the training set; and using the trained metric learning model to control the dynamic mechanical system or another dynamic mechanical system.

A computer-implemented system for fully automating the training, auto-tuning, and deployment of machine learning models at customer sites, especially models for robotic grasping tasks. The system automatically: (1) detects / predicts performance degradation of one or more machine learning models; (2) triggers a new model training / fine-tuning job in response to such detection / prediction; and (3) deploys the new model upon training completion, without stopping or pausing the production line.

Described herein are means for the generation of semantic genesis models through self-supervised learning in the absence of manual labeling, in which the trained semantic genesis models are then utilized for the processing of medical imaging. For instance, an exemplary system is specially configured with means for performing a self-discovery operation which crops 2D patches or crops 3D cubes from similar patient scans received at the system as input; means for transforming each anatomical pattern represented within the cropped 2D patches or the cropped 3D cubes to generate transformed 2D anatomical patterns or transformed 3D anatomical patterns; means for performing a self-classification operation of the transformed anatomical patterns by formulating a C-way multi-class classification task for representation learning; means for performing a self-restoration operation by recovering original anatomical patterns from the transformed 2D patches or transformed 3D cubes having transformed anatomical patterns embedded therein to learn different sets of visual representation; and means for providing a semantics-enriched pre-trained AI model having a trained encoder-decoder structure with skip connections in between based on the performance of the self-discovery operation, the self-classification operation, and the self-restoration operation. Other related embodiments are disclosed.

A computer implemented method includes accessing training data that includes images, associated alternative text, and proximately located text providing instructions describing a user interface. The images are paired with text captions derived from the proximately located text and image's alternative text, training a vision and language model in a self-supervised manner using language masking, region masking, and image-text alignment techniques on respective image region features and tokenized text captions. Fine-tuning of the vision and language model is performed to obtain a specialized model representing user interface elements and associated functions.

The invention discloses an artificial intelligence robot control system fused with a world model architecture, and the system comprises a control main system which comprises a shared multi-mode backbone network module, a strategy head, and a world model head. A visual encoder, an ontology perception encoder, a text encoder and multi-mode fusion are integrated in the shared multi-mode backbone network module, and the strategy head is used for generating a current action instruction. According to the method, a self-supervision signal provided by a world model is used as an additional training constraint, the dependence on large-scale teaching data is reduced, and a strategy head generates actions based on representation rich in physical dynamic information, so that the device has the advantages that the decision is more stable when facing environmental noise or uncertainty; error accumulation in a long-range task is remarkably reduced, and compared with a traditional open-loop strategy model, the method has the advantage that the generalization ability of training out-of-distribution scenes is remarkably improved.

A multi-agent system comprising an administrator configured to maintain a directed acyclic graph, hereinafter DAG, and a plurality of autonomous robots using asymmetric encryption for identification to at least one of the administrator other robots of the plurality of autonomous robots and configured for:a robot of the a plurality of autonomous robots to obtain a physical task from a dynamic task list, to perform the physical task and to, upon completion of the task, append task completion information to the DAG; andanother robot of the a plurality of autonomous robots to, independent of the autonomous robots that has appended the completion information to the DAG, physically and cryptographically verify that a task appended to the DAG by the autonomous robot that has appended the completion information to the DAG has been completed in accordance with the completion information, wherein cryptographic verification includes checking a cryptographic identifier of the autonomous robots that has appended the completion information to the DAG.

The embodiment of the invention provides a multi-regularization self-supervised training method and system for an emotion recognition model. The method comprises the following steps: inputting an electroencephalogram sample into a first emotion recognition model based on masking channel modeling of time spectrum transformation and a second emotion recognition model based on emotion contrast learning; performing emotion recognition on the electroencephalogram sample by using a first emotion recognition model and a second emotion recognition model to obtain a predicted emotion classification result and emotion classification loss; determining a logic value of the electroencephalogram sample through an encoder of the first emotion recognition model and a classifier head, and determining an emotion feature and regularization loss of the electroencephalogram sample through an encoder of the second emotion recognition model and a projection head; and performing self-supervision training based on the emotion classification loss and the regularization loss. According to the multi-regularization self-supervised training in the embodiment of the invention, time-frequency spectrum transformation masking channel modeling and emotion comparison learning pre-training are adopted respectively, so that more accurate emotion information can be captured.

An incremental learning algorithm, extending Rapid Class Augmentation (“XRCA”), implements an unconstrained, recursive least-squares (RLS) style of optimization that incorporates knowledge of all the past training examples into each optimization step by recursively computing an IFCM in a single multi-class prediction head. The single multi-class prediction head receives class token feature vectors from a pretrained, self-supervised transformer model and is able to achieve the same optimal performance as a non-incrementally trained classifier in a jointly optimal manner over a set of increasing classes.

An apparatus to facilitate learning reliable keypoints in situ with introspective self-supervision is disclosed. The apparatus includes one or more processors to provide a view-overlapped keyframe pair from a pose graph that is generated by a visual simultaneous localization and mapping (VSLAM) process executed by the one or more processors; determine a keypoint match from the view-overlapped keyframe pair based on a keypoint detection and matching process, the keypoint match corresponding to a keypoint; calculate an inverse reliability score based on matched pixels corresponding to the keypoint match in the view-overlapped keyframe pair; identify a supervision signal associated with the keypoint match, the supervision signal comprising a keypoint reliability score of the keypoint based on a final pose output of the VSLAM process; and train a keypoint detection neural network using the keypoint match, the inverse reliability score, and the keypoint reliability score.

Methods and systems disclosed herein relate generally to systems and methods for generating visual relationship graphs that identify relationships between objects depicted in an image. A vision-language application uses transformer encoders to generate a graph structure, in which the graph structure represents a dependency between a first region and a second region of an image. The dependency indicates that a contextual representation of the first region was derived, at least in part, by processing the second region. The contextual representation identifies a predicted identity of an image object depicted in the first region. The predicted identity is determined at least in part by identifying a relationship between the first region and other data objects associated with various modalities.

A method of deep structure detection of deep structure signals for a critical dimension scanning electron microscope (CDSEM) image by dividing the image into separate upper and lower regions using self-tunable masking and then denoising and enhancing the upper and lower regions separately, wherein the lower deep structure region is denoised by applying blind denoising by self-supervision (BBSS) thereto.

The present invention discloses a self-supervised group detection method and device for large-scale scenes and multiple objects. The method includes: modeling N people in each t-frame video of a crowded multi-person scene; extracting feature information such as activity and time trajectory for each person; training an environmental perception human behavior simulator in a self-supervised manner, that is, spontaneously learning natural human behavior in a group; based on the idea of ​​causal reasoning, using the environmental perception human behavior simulator to train a relationship network to discover paired interpersonal relationships between multiple people; determining the parameters of the relationship network, and adding a network based on a fully connected layer to realize the task of artificial group detection. The device includes: a processor and a memory. Based on the idea of ​​causal reasoning, the present invention can spontaneously learn natural human behavior in a group without the need for a large amount of labeled information, thereby completing a better group detection task.

A learning device 10 includes a first learning unit 20. The first learning unit 20 includes a first supervised learning unit 22 and a first self-supervised learning unit 24. The first supervised learning unit 22 learns a first object detection network 30 using learning data 40 so as to reduce a first loss between an output of the first object detection network 30 for detecting an object from target image data and supervised data 40B. Using image data 40A and self-supervised data 40C generated from the image data 40A, the first self-supervised learning unit 24 learns the first object detection network 30 so as to reduce a second loss of a feature amount of a corresponding candidate area P between the image data 40A and the self-supervised data 40C, the second loss being derived by the first object detection network 30.

To execute self-supervised learning capable of accepting a generation strategy of various instances for tabular data in which data of heterogeneous formats such as numeric data, categorical data or the like are mixed.SOLUTION: A learning apparatus is a learning apparatus for constructing a machine learning model for tabular format data, and includes: an arithmetic device that executes predetermined processing; a storage device that can be accessed by the arithmetic device; an upstream task learning unit in which the arithmetic device generates an upstream model relating to an upstream task by self-supervised learning based on neighborhood definition information defining a neighborhood relation of a pair of instances of the tabular format data; and a downstream task learning unit in which the arithmetic device generates a downstream model relating to a downstream task by supervised learning, reinforcement learning or unsupervised learning.SELECTED DRAWING: Figure 1

This invention discloses a self-supervised point cloud upsampling method and system based on adversarial learning, comprising the following steps: data acquisition; hybrid geometry-aware downsampling, combining random sampling and farthest point sampling to construct self-supervised pairs conforming to real-world distribution; initial upsampling, achieving global-local feature balance through three-stage iterative refinement; further upsampling, repeating the initial upsampling process within intermediate point clouds and introducing uniformity loss statistics; a detail module, extracting multi-scale geometric features stepwise through three dynamically updated EdgeConv layers and employing a GAN-based adversarial training strategy to output a fine point cloud; a detail-aware discriminator, receiving the predicted fine point cloud and the corresponding ground real point cloud as input, reshaping them into a form suitable for convolutional layers; and point cloud reconstruction, regressing the 3D point cloud shape from the point cloud feature information. Applying this invention can improve the quality and efficiency of high-precision reconstruction for different scenes.

The application provides a spatial transcriptome double self-supervision clustering analysis method and system, relates to the field of bioinformatics, and performs pretreatment on spatial transcriptome data to be analyzed to obtain sample data composed of gene expression data, cell image features and an adjacency matrix; the sample data is input into a trained double self-supervision model to perform clustering, and the clustering distribution of the sample data is obtained; according to the clustering distribution, a cell label is obtained as a final recognized cell type; the double self-supervision is self-supervision training of a linear autoencoder and training of a global target distribution supervised graph convolutional neural network encoder; through self-supervision training of the linear autoencoder and training of the global target distribution supervised graph convolutional neural network encoder, the training effect of the encoder is improved in a double self-supervision manner, and efficient and accurate classification and recognition of spatial transcriptome cells are realized.

A method includes receiving, by a first encoder, an original speech segment, receiving, by a second encoder, an augmented speech segment of the original speech segment, generating, by the first encoder, a first speaker representation based on the original speech segment, generating, by the second encoder, a second speaker representation based on the augmented speech segment, and generating a contrastive loss based on the first speaker representation and the second speaker representation.

A method of deep structure detection of deep structure signals for a critical dimension scanning electron microscope (CDSEM) image by dividing the image into separate upper and lower regions using self-tunable masking and then denoising and enhancing the upper and lower regions separately, wherein the lower deep structure region is denoised by applying blind denoising by self-supervision (BBSS) thereto.

In an embodiment, there is provided a self-supervised representation learning (SSRL) circuitry. The SSRL circuitry includes a normalizer circuitry, and a loss function circuitry. The normalizer circuitry is configured to receive a number, T, batches of embedding features. Each batch includes a number, N, embedding features. The number N corresponds to a number of input samples in a training batch. The number T corresponds to a number of respective transformed batches. Each transformed batch corresponds to a respective transformation of the training batch. The embedding features may be related to the transformed batches. Each embedding feature has a dimension, D, and each embedding feature element corresponds to a respective feature variable. The normalizer circuitry is further configured to normalize each feature variable of a selected batch, using a zero mean and a unit standard deviation of the selected batch. A loss function circuitry is configured to determine a loss function based, at least in part, on a factorizable mixed moment of a plurality of normalized feature variables. The mixed moment is of order K. K is less than or equal to the embedding feature dimension D.

The invention discloses a self-encoding graph convolutional network clustering method based on cross attention fusion, and relates to the field of computer machine learning. According to the invention, a cross attention fusion module is innovatively connected with a content automatic encoder module (CAE), a graph convolution automatic encoder module (GAE) related to single data and a graph convolution automatic encoder module related to the relationship between data are carried out in a layer-by-layer manner, so that discrimination information for a clustering task is highlighted; the module fuses two kinds of heterogeneous numbers, and the self-supervision module restrains distribution represented by middle layers of the CAE and the GAE to be consistent. Experimental results on different types of data sets prove the superiority and robustness of the CaEGCN provided by the invention.

A device, method, and non-transitory computer readable medium that for analyzing computer source code to detect code smells is disclosed. The method includes inputting, via processing circuitry, the source code and creating, via the processing circuitry, pseudo labels by a proxy task based on a vector of tokens for the source code. In addition, the method includes training, via the processing circuitry, a transformer model on the pseudo labels, as a pre-trained model that outputs a prediction of a value of tokens in the vector of tokens, and applying, via the processing circuitry, the pre-trained model to a plurality of fine-tuning models for respective downstream tasks, where each fine-tuning model is created by training the pre-trained model. The method also includes outputting, via the processing circuitry, from each fine-tuning model, an indication of whether a code smell has been detected in the source code.

One or more computer processors obtain an initial subnetwork at a target sparsity and an initial pruning mask from a pre-trained self-supervised learning (SSL) speech model. The one or more computer processors finetune the initial subnetwork, comprising: the one or more computer processors zero out one or more masked weights in the initial subnetwork specified by the initial pruning mask; the one or more computer processors train a new subnetwork from the zeroed out subnetwork; the one or more computer processors prune one or more weights of lowest magnitude in the new subnetwork regardless of network structure to satisfy the target sparsity. The one or more computer processors classify an audio segment with the finetuned subnetwork.

Features from historical transaction and customer data of a financial institution are calculated and / or extracted. Each customer is assigned to a given profitability cluster within each interval of time over a historical period of time based on the corresponding features. A self-supervised machine learning model is trained on the features to predict the clusters in a future interval of time. Features for a most-recent past interval of time are provided as input to the model and the model returns a predicted cluster for a given customer in a future interval of time. When the customer-assigned cluster in the most-recent past interval of time is a higher prioritized cluster than the predicted cluster for the future interval of time, a system of a financial institution (FI) is notified to take one or more mitigating in an attempt to prevent customer churn with the FI.