Wyniki wyszukiwania - BazTech

Ograniczanie wyników

2 2025

Powiadomienia systemowe

Sesja wygasła!
Sesja wygasła!

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: sieć transformerowa

Sortuj według:

Ogranicz wyniki do:

Cloze-tests generation for foreign language learning using transformer networks

Jędrzejewski Sebastian, Szmurło Robert

Przegląd Elektrotechniczny

2025

R. 101, nr 3

247--250

The aim of the paper is to evaluate the transformers network LLM for automatic generation of tests for English language learners with classical approaches. The kind of test we investigate is referred to as cloze test, which are paragraphs of text with gaps which should be filled in by the learners. In the paper, we compare recurrent neural networks to transformer networks (BERT and ELECTRA). Additionally the authors make the training and testing datasets available publicly. The approach related to application of LLMs is based on paper by Felice at al. [2]. In the presented research we extend the loss function and apply extra metrics based on Kullback-Leibler Divergence Loss to improve space distribution of the gaps.

Celem artykułu jest zbadanie możliwości sieci transformerowych LLM do automatycznego generowania testów dla osób uczących się języka angielskiego oraz porównanie wyników z metodami klasycznymi. W artykule badany jest rodzaj testu, który składa się z akapitów tekstu z lukami, które powinni wypełnić uczący się. W artykule porównane są wyniki uzyskiwane za pomocą sieci rekurencyjnych neuronowych oraz sieci transformerowych (BERT i ELECTRA). Podejście związane z zastosowaniem sieci LLM opiera się na artykule Felice i in. [2]. W przedstawionych badaniach autorzy rozszerzają funkcję straty o dodatkowe metryki oparte na stracie dywergencji Kullbacka-Leiblera w celu poprawy rozkładu przestrzennego luk.

Identification of strong motion record baseline drift based on Bayesian-optimized Transformer network

Zhou Baofeng, Yin Yue, Wang Maofa, Zhang Runjie, Zhang Yue, Guo Wenheng

Acta Geophysica

2025

Vol. 73, no. 1

517--525

Research in earthquake engineering heavily relies on strong motion observation. The quality of strong motion records directly affects the reliability of earthquake disaster prevention, rapid reporting of seismic magnitude, earthquake early warning, and other areas. Currently, basic mathematical methods, such as zero-line adjustment and filtering, are commonly employed to ensure the quality of strong motion records. However, these methods often rely on subjective judgment based on human experience when dealing with abnormal waveforms in strong motion records, leading to relatively low efficiency. To address this challenge, this paper proposes an innovative Transformer model based on Bayesian optimization to efficiently identify baseline drift anomalies in strong motion records. By partitioning the strong motion record data from the 1999 Chi- Chi earthquake in Taiwan, China, into two categories: high-quality records (with minimal baseline drift) and low-quality records (with significant baseline drift), we extracted data with distinct features and inputted them into the proposed model for training. Data with distinct features were extracted and input into the proposed model for training. Finally, the model was used to predict whether strong motion records exhibited baseline drift abnormalities. The experimental results show that the optimized Transformer model achieves a performance exceeding 85% in key evaluation metrics such as accuracy and F1 scores. It is capable of efficiently identifying a substantial volume of strong motion records with baseline drift within a short period of time. The model effectively performs the baseline drift classification task for strong motion records and can be used for subsequent identification of abnormalities after baseline drift correction, enabling automation in handling abnormal data related to baseline drift.