International Business Machines Corporation

Xiang Yu Yang

Zhong Fang Yuan

Deng Xin Luo

Tong Liu

Yong Wang

Wen Jie Wang

Ye Wy Wang

Yu Pan

Ci-Wei Lan

Li Juan Gao

Na Liu

Chen Gao

Miao Guo

Jing Du

Chao Zhang

Yi Chen Zhong

De Shuo Kong

Jing Wu

Der-Joung Wang

Yi Ming Wang

Hai B Zou

Kuo-Liang Chou

Si Tong Zhao

Chen Yu Chang

Guan Chao Li

Xin Fang Hao

Min Luo

Ziqiumin Wang

Tzu-Chen Chao

Han Ying Song

Li Jia Lu

Qiang He

Zhi Peng Jia

Wen Jie Hao

Zhi Cao

Rong Fu He

Wang Hu Dang

Jia Yong Xie

Jun Guo

Jia Wei He

Qin Dong Yin

Zhi Yong Jia

Yu Ying Y Y Wang

Yan Yu Zhang

Yu Ying Yy Wang

Han Qiao Yu

Chi Nan

Lin Feng

Provided are a computer program product, system, and method for training and using a vector encoder to determine vectors for sub-images of text in an image to subject to optical character recognition. A vector encoder is trained to encode images representing text into vectors in a vector space. Vectors of images representing similar text have a high degree of cohesion in the vector space. Vectors of images representing dissimilar text have a low degree of cohesion in the vector space. An input image is processed to determine sub-images of the input image that bound text represented in the input image. The sub-images are inputted to the vector encoder to output sub-image vectors. The vector encoder generates a search vector for search text. Optical character recognition is applied to at least one region of the input image including the sub-images having sub-image vectors matching the search vector.

Training and using a vector encoder to determine vectors for sub-images of text in an image subject to optical character recognition

Mechanisms are provided for automatically predicting vehicle weight of a moving vehicle. Vehicle operation data is obtained from one or more on-board sensors/systems of the vehicle and features are extracted. The features are filtered based on a required working condition of the vehicle to identify intervals of features having valid feature data to thereby generate a filtered valid data. The filtered valid data is processed by a first artificial intelligence (AI) computer model based on time slices of the filtered valid data to generate a first prediction of vehicle weight, and by a second AI computer model based on buckets of key variables to generate a second prediction of vehicle weight. The first prediction is fused with the second prediction to generate a final prediction of vehicle weight which is output to downstream computing logic.

Estimating moving vehicle weight based on driving data

Embodiments of the present invention provide an approach for compressing data, and more particularly, to large-scale text data encoding and compression using absolute overfitting on pre-trained language models. Large-scale data is parsed into sentences. A unique token is generated for each sentence to form a token list. A generative (or compression) model is trained from the tokens in the token list to produce the corresponding sentence of each token through absolute overfitting of a pre-trained language model. The compressed text data is stored as the token list and generative model, resulting in a storage space savings.

Large-scale text data encoding and compression

From a set of point data, a set of scattered rays is constructed. From the set of scattered rays, a set of ray slopes is computed. The set of ray slopes is mapped to a corresponding set of trigonometric functions. Using an optimization method, a parameter of the set of trigonometric functions is selected. Using an inverse of the set trigonometric functions, a vehicle mass corresponding to the set of point data is computed. Based on the vehicle mass, a threshold braking distance of a collision avoidance system of the vehicle is adjusted, the threshold braking distance comprising a distance from an object predicted to collide with the vehicle. By braking the vehicle at least the threshold braking distance from the object, a predicted collision between the vehicle and the object is avoided.

Vehicle mass measurement for automated braking

Information extraction and image restructuring includes generating semantic vectors to encode portions of text extracted from a document. For each semantic vector a semantic similarity between a schema key and other text encoded therein is determined based on their respective positions within the document. An enhanced NLP model is created using the semantic vectors, each labeled according to the semantic similarity. The text, including schema key, are re-encoded as a key and candidate vectors. Key-value pairs are generated by matching the key vector with a predetermined number of candidate vectors. The enhanced NLP model, using prompt learning, is repurposed to perform a next-sentence prediction that predicts which of the candidate vectors is logically related to the schema key. Based on the next-sentence prediction, the discrete portion of text identified as the schema key and portion of text determined to be logically related thereto are output.

Information extraction and image re-ordering using prompt learning and machine-reading comprehension

A given segment of computer code is obtained and modified to produce one or more inefficient versions of the given segment of computer code in comparison to the given segment of computer code. A code parse tree is generated for the given segment of computer code and each inefficient version of the given segment of computer code. Model embeddings are generated based on the generated code parse trees and a diffusion model is trained based on the generated model embeddings.

Code performance optimization based on an information-enhanced diffusion model

Reducing an average give-away rate of a weighing device by determining a weight of a product of a weighing device that includes an article, determining one or more conditions of an environment of the weighing device, determining a state of the environment of the weighing device, wherein the state relates to an average give-away rate of the environment of the weighing device, determining a reward value for the state of the environment of the weighing device, wherein the reward value is based at least in part on the weight of the product, and generating a set of parameters for the weighing device based at least in part on the environment, the state, and the reward.

Adjusting parameters of weighing device for reducing average giveaway rate when packaging an article

In an approach, a processor obtains a configuration file of a distributed file system federation, the configuration file comprising a list of a plurality of subclusters within the distributed file system federation and migration trigger factors for the plurality of subclusters. A processor determines a list of one or more source subclusters and a list of to-be-migrated directories in the one or more source subclusters based on a scanning result of the plurality of subclusters and the migration trigger factors in the configuration file. A processor generates a migration plan to migrate the to-be-migrated directories from the one or more source subclusters to one or more target subclusters in the distributed file system federation.

Data migration in a distributed file system

An approach is provided for optimizing a feedback-type question answering process. A training set is constructed to detect missing information of a question. A natural language generation model is trained using the missing information. The natural language generation model is executed to generate a rhetorical question. A response to the rhetorical question is combined with the question to generate an input to a language processor. A new question is generated. The new question is applied to a document library. A final answer is generated.

Question answering information completion using machine reading comprehension-based process

A method, computer program product, and computer system for repairing a Dockerfile. Library versions containing initial version numbers of libraries are extracted from the Dockerfile. A Monte Carlo tree search (MCTS) is executed, using the extracted library versions as input, which generates a tree that includes multiple levels populated with noses. Each node in a level represents the generic library name of a library version in the Dockrerfile and an associated randomly selected version number. At least one of the randomly selected version numbers associated with at least one node in a level differs from the initial version number associated with a versionf. A best successful installation path is selected from the at least one successful installation path. The Dockerfile is repaired by inserting randomly selected version numbers into Dockerfile as replacements for some of the initial version numbers.

Modification of a Dockerfile to repair vulnerabilities existing in the Dockerfile

A method, system, and computer program product for implementing automated pressure level detection and correction is provided. The method includes retrieving from sensors, measurement attributes associated with geological conditions occurring during a mining process. Subsequently, threshold levels configured to activate an alarm associated with measurement attributes exceeding the threshold levels are determined and code is executed with respect to the threshold levels. A combined threshold severity level associated with the safety threshold levels is determined and code is executed with respect to the safety threshold levels. A combined clustering level associated with the safety threshold levels is determined and a difference value between the combined threshold severity level and combined clustering level is generated. A relationship between the difference value and a threshold value is determined and automated software and hardware control systems are enabled for controlling machinery associated with the mining process resulting in operation of the machinery.

Automated pressure level detection and correction

A computer-implemented method, according to one embodiment, includes: receiving an image having characters that correspond to a language, and using a text recognition algorithm to determine a first language believed to correspond to the characters. A first confidence level associated with the first language is also computed, and a determination is made as to whether the first confidence level associated with the first language is outside a predetermined range. In response to determining that the first confidence level associated with the first language is not outside the predetermined range, the first language is output as the given language. The text recognition algorithm is trained using a simple shallow neural network and a generated mixed language corpus. The generated mixed language corpus is formed by: randomly sampling libraries having vocabulary and/or characters therein, and combining the randomly sampled vocabulary and/or characters to form the generated mixed language corpus.

Rapid language detection for characters in images of documents

A method, computer program product, and computer system for generating and using a basic state layer. N task models are provided (N≥2). Each task model was trained on a same pre-trained backbone model. Each task model includes M feature layers and a task layer (M≥1). Each feature layer of each task model includes a parameter matrix that is different for the different models. An encoder-decoder model is trained. The encoder-decoder model includes sequentially: an input layer, an encoder, M hidden layers, a decoder, and an output layer. The encoder is a neural network that maps and compresses the parameter matrices in the input layer into the M hidden layers, which generates a basic state model. The decoder is a neural network that receives the basic state model as input and generates the output layer to be identical to the input layer.

Parameter redundancy reduction method

An example operation may include one or more of monitoring a plurality of brokers within a cluster to identify current workload attributes of the plurality of brokers, determining a health value of a lead broker within the cluster via execution of a machine learning model on current workload attributes of the lead broker, determining to modify resources assigned to the lead broker based on the determined health value of the lead broker, executing an optimization algorithm on the current workload attributes of the plurality of brokers within the cluster to determine an optimum task distribution, and reallocating tasks amongst the lead broker and the one or more other brokers within the cluster based on the optimum task distribution.

Balanced optimization within a broker cluster

A method, computer program product and computer system to provide topic guide during document drafting is provided. A processor retrieves at least one section of text from a document. A processor receives a target topic for the document. A processor extracts at least one local topic from the at least one section of text. A processor generates a semantic network comprising the at least one local topic and the target topic. A processor determines a deviation value for the at least one local topic based on a distance between the at least one local topic and the target topic in the semantic network. A processor, in response to the deviation value exceeding a threshold value, alerts a user that the at least one section of text from the document is off-topic from the target topic.

Computer-assisted topic guidance in document writing

Growing community of inventors

Xi'an, China

Xiang Yu Yang