Vianai Systems, Inc.

Sap Se

Sap Ag

Vishal Inder Sikka

Sanjay Rajagopalan

Yoshiki Ohshima

Matthias Gerhard Anlauff

Jun Zhu

Tao Liu

Janaki P Kumar

Bingjun Zhou

Xiao Wang

Kevin Frederick Dunnell

Philipp Maximilian Wesel

Rahim Mohamed Yaseen

Shai Agassi

Asuman Suenbuel

Daniel James Amelang

Walid Rahman

Richard Waldinger

Xiaoqi Ouyang

Samuel J Yen

Kyle Richardson

Jeong H Kim

One embodiment of the present invention sets forth a technique for simplifying a trained machine learning model. The technique includes determining a first set of images associated with a first output class predicted by the trained machine learning model. The technique also includes generating a first logical representation of the first set of images, wherein the first logical representation includes one or more conjunctions of a first set of pixel values included in a first image and a disjunction of the first set of pixel values and a second set of pixel values included in a second image. The technique further includes generating a simplified representation of the trained machine learning model that includes a first mapping of the first logical representation to the first output class, wherein the first mapping indicates that the trained machine learning model predicts the first output class for one or more input images.

Simplifying convolutional neural networks using logical representations of images

Various embodiments set forth systems and techniques for generating domain-specific question answering models. The techniques include receiving a set of input text corresponding to a particular domain; generating, based on the set of input text, a question-answer dataset corresponding to the particular domain, the question-answer dataset comprising a plurality of question-answer pairs; and causing one or more machine learning algorithms to be applied to the question-answer dataset to generate a question answering model associated with the particular domain.

Text to question-answer model system

One embodiment of the present invention sets forth a technique for simplifying a trained machine learning model. The technique includes determining a first set of images associated with a first output class predicted by the trained machine learning model. The technique also includes generating a first aggregated representation of the first set of images, wherein the first aggregated representation includes a first plurality of representative pixel values for a plurality of pixel locations included in the first set of images. The technique further includes generating a simplified representation of the trained machine learning model that includes a first mapping of the first aggregated representation to the first output class, wherein the first mapping indicates that the trained machine learning model predicts the first output class for one or more input images.

Simplifying convolutional neural networks using aggregated representations of images

An artificial intelligence (AI) model includes one or more feature models coupled to one or more observer models in a hierarchical fashion. The feature models are configured to process an input to detect different features within that input. The observer models are configured to analyze the operation of the feature models during processing of the input to generate various types of observations. One type of observation includes a natural language expression that conveys how various architectural and/or functional characteristics of a given feature model influence the processing of the input to detect features, thereby exposing the underlying mechanisms via which the given feature model operates.

Techniques for generating natural language descriptions of neural networks

As described, an artificial intelligence (AI) design application exposes various tools to a user for generating, analyzing, evaluating, and describing neural networks. The AI design application includes a network generator that generates and/or updates program code that defines a neural network based on user interactions with a graphical depiction of the network architecture. The AI design application also includes a network analyzer that analyzes the behavior of the neural network at the layer level, neuron level, and weight level in response to test inputs. The AI design application further includes a network evaluator that performs a comprehensive evaluation of the neural network across a range of sample of training data. Finally, the AI design application includes a network descriptor that articulates the behavior of the neural network in natural language and constrains that behavior according to a set of rules.

Techniques for creating, analyzing, and modifying neural networks

Techniques for visualizing the operation of neural networks using samples of training data

Techniques for modifying the operation of neural networks

An artificial intelligence (AI) design application that exposes various tools to a user for generating, analyzing, evaluating, and describing neural networks. The AI design application includes a network generator that generates and/or updates program code that defines a neural network based on user interactions with a graphical depiction of the network architecture. The network generator enables a developer to define the neural network architecture using a pipeline of mathematical expressions that can be directly compiled without the need of a complex software stack. The compilation process allows for the variables to be learned during the training process to be left unassigned when the neural network is instantiated. In particular, the compiler identifies such unassigned variables as variables having values that will be determined during the training process.

Techniques for defining and executing program code specifying neural network architectures

Systems and methods are presented for performing natural language processing and reasoning. In some embodiments, a computer-implemented methods is presented. The method may include accessing a natural language query from a user, parsing the natural language query into a computer-interpretable semantic representation, converting the semantic representation into a computer-interpretable logical syntax, determining a solution to the computer-interpretable logical syntax using a reasoning engine and at least one data source, and outputting an answer to the natural language query using the solution to the computer-interpretable logical syntax.

Systems and methods for natural language processing using machine-oriented inference rules

Techniques are described for interacting with a nature inspired virtual environment to present information to the user. The information can be presented to the user by using a collection of icons in the virtual environment. The virtual environment and the collection of icons can resemble a scene from nature. As a result, the additional information is presented to the user without disrupting the visual appearance of the nature inspired virtual environment.

Search in a nature inspired user interface

A system is described for providing a virtual environment corresponding to a scene from nature. The virtual environment can be used to present performance metrics to a user. In some examples, the virtual environment can be a pond where performance metrics are represented using lily pads. The visual appearance of the lily pad can be dependent on metadata of the corresponding performance metric and the lily pads can be grouped according to how the performance metrics are grouped in the database.

Nature inspired interaction paradigm

A business management system for visualizing transactional data objects in real time is provided. An example system accesses a stream of transactional data objects and generates a in a three-dimensional graphical paradigm. A viewer may manipulate the presentation of the transactional data objects by engaging gestures and visual controls that may be provided on a display screen.

Real-time visualization of transactional data objects

To enable language-level integration of multiple programming languages into a single programming framework, one or more domain specific languages (DSLs) are specified as incremental extensions to a host programming language. An application created in this framework may include program code written in the host language as well as DSL inset code that is written in any of the DSLs. The extensions to the host language are described in DSL specifications, which define how the DSL inset code is processed. At runtime, DSL inset code in the source code is detected and passed to a DSL inset processor for execution according to an appropriate DSL specification, whereas source code in the host language is executed by a host language processor.

Execution of program code having language-level integration of program models

Embodiments of the present invention include the creation of spreadsheet software services. In one embodiment, the features and functionality of a spreadsheet may be transformed into a software service, such as a web service, for example. In one embodiment, the present invention includes a method of generating a software service comprising analyzing a spreadsheet to extract an output schema corresponding to at least a portion of the spreadsheet, generating a software service interface description comprising the output schema and the description of the output data, wherein the software service interface is accessible over a network, and accessing the functionality of the spreadsheet using the software service interface description.

Growing community of inventors

Los Altos, CA, United States of America

Vishal Inder Sikka