Adobe, Inc.

Hung Hai Bui

Trung Huu Bui

Branislav Kveton

Michael F Kraley

Jaya Kawale

Nikolaos Vlassis

Pranjal Daga

Hailin Jin

Eunyee Koh

Quan Anh Tran

Sungchul Kim

Atanu Ranjan Sinha

Shawn Alan Gaither

Walter Wei-Tuh Chang

Mohammad Ghavamzadeh

Nikhil Sheoran

Deepali Jain

Deepali Gupta

Charles Li Chen

Suvash Sedhain

Anirban Roychowdhury

Walter Chang

Franck Dernoncourt

Georgios Theocharous

Trevor Paulsen

Sheng Li

Tuan Manh Lai

Craig Mathis

Ivan Andrus

John Thomas Kucera

Kristopher Paries

Michael Rimer

Quan Hung Tran

Hamid Dadkhahi

Chunyuan Li

The present description relates to systems, methods, and non-transitory computer readable media for generating digital responses to digital queries by utilizing a classification model and query-specific analysis models. For example, the described systems can train a classification model to generate query classifications corresponding to product queries, conversational queries, and/or recommendation/purchase queries. Moreover, the described systems can apply the classification model to select pertinent models for particular queries. For example, upon classifying a product query, the described systems can utilize a neural ranking model (trained based on a set of training product specifications and training queries) to generate relevance scores for product specifications associated with a digital query. The described systems can further compare generated relevance scores to select a product specification and generate a digital response that includes the pertinent product specification to provide for display to a client device.

Generating and utilizing classification and query-specific models to generate digital responses to queries from client device

The present invention is directed towards providing automated workflows for the identification of a reading order from text segments extracted from a document. Ordering the text segments is based on trained natural language models. In some embodiments, the workflows are enabled to perform a method for identifying a sequence associated with a portable document. The methods includes iteratively generating a probabilistic language model, receiving the portable document, and selectively extracting features (such as but not limited to text segments) from the document. The method may generate pairs of features (or feature pair from the extracted features). The method may further generate a score for each of the pairs based on the probabilistic language model and determine an order to features based on the scores. The method may provide the extracted features in the determined order.

Probabilistic language models for identifying sequential reading order of discontinuous text segments

Digital content interaction prediction and training techniques that address imbalanced classes are described. In one or more implementations, a digital medium environment is described to predict user interaction with digital content that addresses an imbalance of numbers included in first and second classes in training data used to train a model using machine learning. The training data is received that describes the first class and the second class. A model is trained using machine learning. The training includes sampling the training data to include at least one subset of the training data from the first class and at least one subset of the training data from the second class. Iterative selections are made of a batch from the sampled training data. The iteratively selected batches are iteratively processed by a classifier implemented using machine learning to train the model.

Digital content interaction prediction and training that addresses imbalanced classes

Metric forecasting techniques and systems in a digital medium environment are described that leverage similarity of elements, one to another, in order to generate a forecast value for a metric for a particular element. In one example, training data is received that describes a time series of values of the metric for a plurality of elements. The model is trained to generate the forecast value of the metric, the training using machine learning of a neural network based on the training data. The training includes generating dimensional-transformation data configured to transform the training data into a simplified representation to determine similarity of the plurality of elements, one to another, with respect to the metric over the time series. The training also includes generating model parameters of the neural network based on the simplified representation to generate the forecast value of the metric.

Metric forecasting employing a similarity determination in a digital medium environment

Techniques disclosed herein provide more efficient and more relevant item recommendations to users in large-scale environments in which only positive interest information is known. The techniques use a rank-constrained formulation that generalizes relationships based on known user interests in items and/or use a randomized singular value decomposition (SVD) approximation technique to solve the formulation to identify items of interest to users in an efficiently, scalable manner.

Item recommendation techniques

The present disclosure relates to systems, methods, and non-transitory computer readable media for generating dialogue responses based on received utterances utilizing an independent gate context-dependent additive recurrent neural network. For example, the disclosed systems can utilize a neural network model to generate a dialogue history vector based on received utterances and can use the dialogue history vector to generate a dialogue response. The independent gate context-dependent additive recurrent neural network can remove local context to reduce computation complexity and allow for gates at all time steps to be computed in parallel. The independent gate context-dependent additive recurrent neural network maintains the sequential nature of a recurrent neural network using the hidden vector output.

Generating dialogue responses utilizing an independent context-dependent additive recurrent neural network

Systems and methods are disclosed herein for providing a user interface representing differences between segments of end users. The systems and methods receive user input on a user interface identifying a first segment, the first segment being a subset of the end users having a particular characteristic, determine differences between the first segment and a second segment, and represent, on the user interface, the differences between the first segment and the second segment based on relative significances of the differences. The marketer using the user interface is able to quickly and easily identify the metrics, dimensions, and/or relationships to other segments that most distinguish the compared segments from one another.

Audience comparison

Certain embodiments involve determining and outputting correlations between metrics in large-scale web analytics datasets. For example, a processor identifies pairs of data metrics in a web analytics data set and determines a Maximal Information Coefficient (MIC) score for each pair of data metrics that indicates a strength of a correlation between the pair of data metrics. The processor generates an interactive user interface that graphically displays each pair of correlated data metrics having an MIC score above a threshold and the interactive user interface indicates the strength of the correlation between each displayed pair of correlated data metrics. The processor receives user input indicating an adjustment to the threshold and modifies the interactive user interface in response to receiving the user input by adding pairs of correlated data metrics to, or removing pairs of correlated metrics from, the interactive user interface based on the adjustment to the threshold.

Methods and systems for determining and outputting correlations between metrics in a web analytic dataset

Generating dialogue responses in end-to-end dialogue systems utilizing a context-dependent additive recurrent neural network

Techniques for predictively selecting a content presentation in a client-server computing environment are described. In an example, a content management system detects an interaction of a client with a server and accesses client features. Responses of the client to potential content presentations are predicted based on a multi-task neural network. The client features are mapped to input nodes and the potential content presentations are associated with tasks mapped to output nodes of the multi-task neural network. The tasks specify usages of the potential content presentations in response to the interaction with the server. In an example, the content management system selects the content presentation from the potential content presentations based on the predicted responses. For instance, the content presentation is selected based on having the highest likelihood. The content management system provides the content presentation to the client based on the task corresponding to the content presentation.

Content presentation based on a multi-task neural network

Systems and methods provide for generating predictive models that are useful in predicting next-user-actions. User-specific navigation sequences are obtained, the navigation sequences representing temporally-related series of actions performed by users during navigation sessions. To each navigation sequence, a Recurrent Neural Network (RNN) is applied to encode the navigation sequences into user embeddings that reflect time-based, sequential navigation patterns for the user. Once a set of navigation sequences is encoded to a set of user embeddings, a variety of classifiers (prediction models) may be applied to the user embeddings to predict what a probable next-user-action may be and/or the likelihood that the next-user-action will be a desired target action.

Predictive analysis of target behaviors utilizing RNN-based user embeddings

Automated workflows for identification of reading order from text segments using probabilistic language models

Certain embodiments identify a correct structured reading-order sequence of text segments extracted from a file. A probabilistic language model is generated from a large text corpus to comprise observed word sequence patterns for a given language. The language model measures whether splicing together a first text segment with another continuation text segment results in a phrase that is more likely than a phrase resulting from splicing together the first text segment with other continuation text segments. Sets of text segments, which include a first set with a first text segment and a first continuation text segment as well as a second set with the first text segment and a second continuation text segment, are provided to the probabilistic model. A score indicative of a likelihood of the set providing a correct structured reading-order sequence is obtained for each set of text segments.

Identification of reading order text segments with a probabilistic language model

Growing community of inventors

Sunnyvale, CA, United States of America

Hung Hai Bui