Adobe, Inc.

Ritwik Sinha

Viswanathan Swaminathan

David Arbour

Moumita Sinha

Md Mehrab Tanjim

Saayan Mitra

William Brandon George

Shiv Kumar Saini

Krishna Kumar Singh

Somdeb Sarkhel

Niranjan Shivanand Kumbi

Sunny Dhamnani

Sridhar Mahadevan

Ryan A Rossi

Eunyee Koh

Wei Zhang

Sungchul Kim

Atanu Ranjan Sinha

Handong Zhao

John Philip Collomosse

Tung Mai

Trevor Paulsen

Tong Yu

Xiang Chen

Shunan Guo

Gaurav Sinha

Omar Rahman

Said Kobeissi

Trisha Mittal

Simon Jenni

Jordan Walker

Vaidyanathan Venkatraman

Sreekanth Reddy

Prithvi Bhutani

Akangsha Sunil Bedmutha

Kai Yeung Lau

Amit Doda

Yifu Zheng

Alan Lai

Jan Kadel

Tyler Rasmussen

Samuel Nephi Grigg

Abhisek Trivedi

A system debiases image translation models to produce generated images that contain minority attributes. A balanced batch for a minority attribute is created by over-sampling images having the minority attribute from an image dataset. An image translation model is trained using images from the balanced batch by applying supervised contrastive loss to output of an encoder of the image translation model and an auxiliary classifier loss based on predicted attributes in images generated by a decoder of the image translation model. Once trained, the image translation model is used to generate images with the minority image when given an input image having the minority attribute.

Debiasing image to image translation models

Techniques for video manipulation detection are described to detect one or more manipulations present in digital content such as a digital video. A detection system, for instance, receives a frame of a digital video that depicts at least one entity. Coordinates of the frame that correspond to a gaze location of the entity are determined, and the detection system determines whether the coordinates correspond to a portion of an object depicted in the frame to calculate a gaze confidence score. A manipulation score is generated that indicates whether the digital video has been manipulated based on the gaze confidence score. In some examples, the manipulation score is based on at least one additional confidence score.

Video manipulation detection

Techniques for web bot detection using behavioral analysis and machine learning are disclosed. In an example method, a processing device receives an indication of a network interaction by a client agent, from which behaviors of the client agent can be determined. A heuristics module may classify the client agent as in an unknown class based on the behaviors of the client agent. A trained adversarial neural network may also classify the client agent as in the unknown class. The processing device then generates a graph representation of the network interaction. A trained graph convolutional neural network may classify the client agent as in a bot class using the graph representation. Based on the classification of the client agent as a bot, the processing device executes a command to cause a bot countermeasure and generates a notification including information about the behaviors of the client agent.

Web bot detection via behavioral analysis and machine learning

Embodiments are disclosed for performing content authentication. A method of content authentication may include dividing a query video into a plurality of chunks. A feature vector may be generated, using a fingerprinting model, for each chunk from the plurality of chunks. Similar video chunks are identified from a trusted chunk database based on the feature vectors using a multi-chunk search policy. One or more original videos corresponding to the query video are then returned.

Scalable video fingerprinting for content authenticity

In some embodiments, techniques for identifying email events generated by bot activity are provided. For example, a process may involve applying bot detection patterns to identify bot activity among email response events.

Bot activity detection for email tracking

Graphic visualizations, such as charts or graphs conveying data attribute values, can be generated based on natural language queries, i.e., natural language requests. To do so, a natural language request is parsed into n-grams, and from the n-grams, word embeddings are determined using a natural language model. Data attributes for the graphic visualization are discovered in the vector space from the word embeddings. The type of graphic visualization can be determined based on a request intent, which is determined using a trained intent classifier. The graphic visualization is generated to include the data attribute values of the discovered data attributes, and in accordance with the graphic visualization type.

Learning-based method for generating visualizations via natural language

Various disclosed embodiments are directed to using one or more algorithms or models to select a suitable or optimal variation, among multiple variations, of a given content item based on feedback. Such feedback guides the algorithm or model to arrive at suitable variation result such that the variation result is produced as the output for consumption by users. Further, various embodiments resolve tedious manual user input requirements and reduce computing resource consumption, among other things, as described in more detail below.

Media content enhancement based on user feedback of multiple variations

The present disclosure relates to systems, non-transitory computer-readable media, and methods for generating automatic suggestions to effectively modify the organization of an ingested data collection without destruction of the underlying raw data. In particular, in one or more embodiments, the disclosed systems utilize multiple machine learning models in sequence to determine likelihoods that the organizational structure of an ingested data collection should be modified in various ways. In response to generating these likelihoods, the disclosed systems generate corresponding automatic suggestions to modify the organization of the ingested data collection. In response to a detected selection of one or more of the automatic suggestions, the disclosed systems read data out of the ingested data collection in accordance with the selected automatic suggestions to effectively modify the organization of the ingested data collection.

Generating and executing automatic suggestions to modify data of ingested data collections without additional data ingestion

Systems and methods for diversity auditing are described. The systems and methods include identifying a plurality of images; detecting a face in each of the plurality of images using a face detection network; classifying the face in each of the plurality of images based on a sensitive attribute using an image classification network; generating a distribution of the sensitive attribute in the plurality of images based on the classification; and computing a diversity score for the plurality of images based on the distribution.

System and methods for diversity auditing

Methods and systems disclosed herein relate generally to systems and methods for generating simulated images for enhancing socio-demographic diversity. An image-generating application receives a request that includes a set of target socio-demographic attributes. The set of target socio-demographic attributes can define a gender, age, and/or race of a subject that are non-stereotypical for a particular occupation. The image-generating application applies the a machine-learning model to the set of target socio-demographic attributes. The machine-learning model generates a simulated image depicts a subject having visual characteristics that are defined by the set of target socio-demographic attributes.

Generating simulated images that enhance socio-demographic diversity

The present disclosure relates to a feature contribution system that accurately and efficiently provides the influence of features utilized in machine-learning models with respect to observed model results. In particular, the feature contribution system can utilize an observed model result, initial contribution values, and historical feature values to determine a contribution value correction factor. Further, the feature contribution system can apply the correction factor to the initial contribution values to determine correction-factor adjusted contribution values of each feature of the model with respect to the observed model result.

Efficiently determining local machine learning model feature contributions

In implementations of systems for predicting a terminal event, a computing device implements a termination system to receive input data defining a period of time and a maximum event threshold. This system uses a classification model to generate event scores for a plurality of entity devices. Each of the event scores indicates a probability of an event occurrence for a corresponding entity device within a period of time. The plurality of entity devices are segmented into a first segment and a second segment based on an event score threshold. Entity devices included in the first segment have event scores greater than the event score threshold and entity devices included in the second segment have event scores below the event score threshold. The termination system generates an indication of a probability that a number of event occurrences for the entity devices included in the second segment exceeds the maximum even threshold within the period of time.

Systems for predicting a terminal event

This disclosure involves generating graph data structures that model inter-feature dependencies for use with machine-learning models to predict end-user behavior. For example, a processing device receives an input dataset and a request to modify a first input feature of the input dataset. The processing device uses a graph data structure that models the inter-feature dependencies to modify the input dataset by propagating the modification of the first input feature to a second input feature dependent on the first input feature. The modification to the second input feature is a function of at least (a) the value of the first input feature and (b) a weight assigned to an edge linking the first input feature to the second input feature within the directed graph. The processing device then applies a trained machine-learning model to the modified input dataset to generate a prediction of an outcome.

Graph data structure for using inter-feature dependencies in machine-learning

A method, apparatus, and non-transitory computer readable medium for data analytics are described. Embodiments of the method, apparatus, and non-transitory computer readable medium include monitoring online activity corresponding to a plurality of users; receiving aggregate marketing data for a marketing activity; identifying online activity data for a time period corresponding to the marketing activity based on the monitoring; generating a regression model based on the aggregate marketing data and the online activity data using Bayesian regression, wherein the regression model represents a relationship between the marketing activity and the online activity, comprises a time effect coefficient, and is based on a prior distribution of the time effect coefficient that decays to zero as time increases; and estimating a treatment effect for the marketing activity on the online activity based on the regression model, wherein the treatment effect comprises a rate of effect decay.

Bayesian estimation of the effect of aggregate advertising on web metrics

Growing community of inventors

Cupertino, CA, United States of America

Ritwik Sinha