International Business Machines Corporation

Etienne Eben Vos

Francois Pierre Luus

Komminist S Weldemariam

Ismail Yunus Akhalwaya

Naweed Aghmad Khan

Ndivhuwo Makondo

Campbell D Watson

Ryan Nelson Riegel

Sibusisiwe Audrey Makhanya

Zubeida Patel

Thabang Mathonsi

A method, computer program, and computer system are provided for identifying bias in weather models. Data corresponding to one or more forecasts associated with a weather model is received. One or more forecast errors in the received data are identified. A forecast bias is determined from among the one or more forecast errors based on determining a presence of consistent errors in a plurality of regions associated with the received data over a period of time. The weather model is updated based on minimizing the determined forecast bias.

Weather/climate model forecast bias explainability

An ontology topic is selected and a pretrained predictive language model is primed to create a predictive primed model based on one or more ontological rules corresponding to the selected ontology topic. Using the predictive primed model, natural language text is generated based on the ontology topic and guidance of a prediction steering component. The predictive primed model is guided in selecting text that is predicted to be appropriate for the ontology topic and the generated natural language text. The generated natural language text is processed to generate extracted ontology rules and the extracted ontology rules are compared to one or more rules of an ontology rule database that correspond to the ontology topic. A check is performed to determine if a performance of the ontology extractor is acceptable.

Generative ontology learning and natural language processing with predictive language models

Train a machine learning model, using an image-based knowledge graph of tropical cyclone data, for implementing a surface field modeling architecture that produces images of at least surface wind fields and surface rainfall fields from images of at least tropical cyclone tracks and pressure intensities. Generate model images of a modeled surface wind field and a modeled surface rainfall field by providing images of at least a user-generated tropical cyclone track and pressure intensity to the trained machine learning model.

Modeling tropical cyclone surface fields for impact assessment

Methods and systems for generating representative data. A generator is configured to create, using a learning model, one or more generated records based on a plurality of training records obtained from a sensitive database. A discriminator is trained to identify the generated records as being generated based on the training records and a privacy adversary is trained to identify a training sample as being more similar to a distribution of the generated records than a distribution of the reference records.

Generation of representative data to preserve membership privacy

Maximum expressivity can be received representing a ratio between maximum and minimum input weights to a neuron of a neural network implementing a weighted real-valued logic gate. Operator arity can be received associated with the neuron. Logical constraints associated with the weighted real-valued logic gate can be determined in terms of weights associated with inputs to the neuron, a threshold-of-truth, and a neuron threshold for activation. The threshold-of-truth can be determined as a parameter used in an activation function of the neuron, based on solving an activation optimization formulated based on the logical constraints, the activation optimization maximizing a product of expressivity representing a distribution width of input weights to the neuron and gradient quality for the neuron given the operator arity and the maximum expressivity. The neural network of logical neurons can be trained using the activation function at the neuron, the activation function using the determined threshold-of-truth.

Optimizing capacity and learning of weighted real-valued logic

A first set of signal data is received. Generative machine learning models are trained based on the first set of signal data. The generative machine learning models include at least a first model trained to identify a first signal component and a second model trained to identify a second signal component. An incoming mixed signal data stream is dynamically separated into a clean signal component and a noise signal component by running the generative machine learning models.

Growing community of inventors

Johannesburg, South Africa

Etienne Eben Vos