公开(公告)号：US20240346106A1

公开(公告)日：2024-10-17

申请号：US18298727

申请日：2023-04-11

申请人： INTERNATIONAL BUSINESS MACHINES CORPORATION

发明人： Lior Horesh ,  Cristina Cornelio ,  Sanjeeb Dash ,  Joao P. Goncalves ,  Kenneth Lee Clarkson ,  Nimrod Megiddo ,  Bachir El Khadir ,  Vernon Ralph Austel

IPC分类号： G06F17/11

CPC分类号： G06F17/11

摘要： A method for obtaining a refined model given a mis-specified symbolic model. The method includes receiving a mis-specified symbolic model and data pertaining to a process or phenomenon corresponding to the mis-specified symbolic model; receiving one or more constraints; generating a plurality of partial expression trees based on the mis-specified symbolic model; solving an optimization problem for each of the partial expression trees; and determining a refined symbolic model of the mis-specified symbolic model based on results of the optimization problem for each partial expression tree.

公开(公告)号：US20240330710A1

公开(公告)日：2024-10-03

申请号：US18194605

申请日：2023-03-31

申请人： INTERNATIONAL BUSINESS MACHINES CORPORATION

发明人： Lior Horesh ,  Cristina Cornelio ,  Bachir El Khadir ,  Sanjeeb Dash ,  Joao P. Goncalves ,  Kenneth Lee Clarkson

IPC分类号： G06N5/01

CPC分类号： G06N5/013

摘要： A method generates automated discovery of new scientific formulas. The method includes receiving a background theory associated with a phenomenon being studied. The processor receives a set of training data associated with the phenomenon being studied. The set of training data is processed in a machine learning model that generates candidate formulas from data points in the set of training data. Values of a numerical error-vector are generated for the candidate formulas. The candidate formulas are processed in a reasoning model. The operation of the reasoning model includes generating values of a theoretical error-vector based on the background theory. An output of a performance metric is generated based on a generalization of the theoretical error-vector and a reasoning error. The processor determines whether one of the candidate formulas is a meaningful and valid new scientific formula, based on a behavior of the reasoning error and the reasoning performance metric.

公开(公告)号：US20240330535A1

公开(公告)日：2024-10-03

申请号：US18190239

申请日：2023-03-27

申请人： International Business Machines Corporation

发明人： Lior Horesh ,  Bachir El Khadir ,  Sanjeeb Dash ,  Kenneth Lee Clarkson ,  Cristina Cornelio

IPC分类号： G06F30/20

CPC分类号： G06F30/20 ,  G06F2111/02 ,  G06F2111/06

摘要： Embodiments of the invention are directed to a programmable computer system having a processor system operable to perform processor system operations that include representing a set of candidate functions in a mathematical expression domain. The set of candidate functions defines relationships between data of an existing system. A set of known background theory is represented in the mathematical expression domain. The set of known background theory defines known relationships associated with the existing system. A model composition operation is performed that includes analyzing, in the mathematical expression domain, the set of candidate functions and the set of known background theory to generate a composed model that satisfies a target data fidelity in a manner that also satisfies a predetermined level of compatibility between the composed model and the set of known background theory.