Transformer Models for Paraphrase Detection: A Comprehensive Semantic Similarity Study

One hundred and sixty-eight genotypes of cotton from the same growing region were used as a germplasm group to study the validity of different genetic distances in constructing cotton core subset. Mixed linear model approach was employed to unbiasedly predict genotypic values of 20 traits for eliminating the environmental effect. Six commonly used genetic distances (Euclidean, standardized Euclidean, Mahalanobis, city block, cosine and correlation distances) combining four commonly used hierarchical cluster methods (single distance, complete distance, unweighted pair-group average and Ward's methods) were used in the least distance stepwise sampling (LDSS) method for constructing different core subsets. The analyses of variance (ANOVA) of different evaluating parameters showed that the validities of cosine and correlation distances were inferior to those of Euclidean, standardized Euclidean, Mahalanobis and city block distances. Standardized Euclidean distance was slightly more effective than Euclidean, Mahalanobis and city block distances. The principal analysis validated standardized Euclidean distance in the course of constructing practical core subsets. The covariance matrix of accessions might be ill-conditioned when Mahalanobis distance was used to calculate genetic distance at low sampling percentages, which led to bias in small-sized core subset construction. The standardized Euclidean distance is recommended in core subset construction with LDSS method.

This study evaluates the effect of retrieval embeddings and similarity metrics at the generative component (LLaMA3). No fine-tuning or decoding variation was applied to the generator, ensuring that any performance differences were due to retrieval quality alone. This design enables a focused evaluation of how retrieval configurations influence the final output in a controlled RAG setup, particularly for primary school topics in sports and health (PJOK). We point out that RAG is a technique that helps large language models make fewer mistakes by using outside information, which improves the accuracy of facts and the relevance of the context. The research compares various embedding models and retrieval similarity metrics to evaluate their effectiveness in retrieving relevant information. The baai/bge-m3 and gte-multilingual-base models demonstrate strong performance with 0.98 precision, while the indo-bge-m3 model shows high recall but lower precision. From 5 embeddings, Llama Embedding performs significantly worse than the other models. The dot product has higher recall than Euclidean and cosine similarity metrics but embedding model selection significantly influences overall performance. The study underscores the importance of embedding choice and retrieval quality in improving chatbot responses. Further refinements are suggested, including better filtering, reranking retrieved chunks, and fine-tuning the retriever, or LLaMA, to align with the target language domain, such as Bahasa Indonesia. The results show that even though RAG improves how well things are generated, the quality of what is retrieved is still a major problem, which means we need to keep making improvements for the best results in educational uses.

In natural language processing (NLP) tasks, computing semantic textual similarity (STS) is crucial for capturing nuanced semantic differences in text. Traditional word vector methods, such as Word2Vec and GloVe, as well as deep learning models like BERT, face limitations in handling context dependency and polysemy and present challenges in computational resources and real-time processing. To address these issues, this paper introduces two novel methods. First, a sentence embedding generation method based on Kullback–Leibler Divergence (KLD) optimization is proposed, which enhances semantic differentiation between sentence vectors, thereby improving the accuracy of textual similarity computation. Second, this study proposes a framework incorporating RoBERTa knowledge distillation, which integrates the deep semantic insights of the RoBERTa model with prior methodologies to enhance sentence embeddings while preserving computational efficiency. Additionally, the study extends its contributions to sentiment analysis tasks by leveraging the enhanced embeddings for classification. The sentiment analysis experiments, conducted using a Stochastic Gradient Descent (SGD) classifier on the ACL IMDB dataset, demonstrate the effectiveness of the proposed methods, achieving high precision, recall, and F1 score metrics. To further augment model accuracy and efficacy, a feature selection approach is introduced, specifically through the Dynamic Principal Component Selection (DPCS) algorithm. The DPCS method autonomously identifies and prioritizes critical features, thus enriching the expressive capacity of sentence vectors and significantly advancing the accuracy of similarity computations. Experimental results demonstrate that our method outperforms existing methods in semantic similarity computation on the SemEval-2016 dataset. When evaluated using cosine similarity of average vectors, our model achieved a Pearson correlation coefficient (τ) of 0.470, a Spearman correlation coefficient (ρ) of 0.481, and a mean absolute error (MAE) of 2.100. Compared to traditional methods such as Word2Vec, GloVe, and FastText, our method significantly enhances similarity computation accuracy. Using TF-IDF-weighted cosine similarity evaluation, our model achieved a τ of 0.528, ρ of 0.518, and an MAE of 1.343. Additionally, in the cosine similarity assessment leveraging the Dynamic Principal Component Smoothing (DPCS) algorithm, our model achieved a τ of 0.530, ρ of 0.518, and an MAE of 1.320, further demonstrating the method’s effectiveness and precision in handling semantic similarity. These results indicate that our proposed method has high relevance and low error in semantic textual similarity tasks, thereby better capturing subtle semantic differences between texts.

This paper investigates the impact of geometric semantic crossover operators in a wide range of symbolic regression problems. First, it analyses the impact of using Manhattan and Euclidean distance geometric semantic crossovers in the learning process. Then, it proposes two strategies to numerically optimize the crossover mask based on mathematical properties of these operators, instead of simply generating them randomly. An experimental analysis comparing geometric semantic crossovers using Euclidean and Manhattan distances and the proposed strategies is performed in a test bed of twenty datasets. The results show that the use of different distance functions in the semantic geometric crossover has little impact on the test error, and that our optimized crossover masks yield slightly better results. For SGP practitioners, we suggest the use of the semantic crossover based on the Euclidean distance, as it achieved similar results to those obtained by more complex operators.

Clustering is a data mining method that is widely used to group data based on similarity. This clustering process can be used to streamline data so as to facilitate the data ranking process. The purpose of this study was to make comparisons of distance measurements on the K-Means and TOPSIS methods to select students who would take part in industrial visit activities. The method used in this study is the K-Means Algorithm to carry out the clustering process whose results will be processed using the TOPSIS method, both of which use Euclidean, Manhattan and Minkowsky Distance. Based on the clustering process, there were 21 respondents who were eligible to be included, then with TOPSIS a ranking process was carried out. Of the three distance measurements used based on the Pearson Euclidean distance correlation test, the highest results were 0.992, Manhattan 0.982 and Minkowsky 0.980, with ratings one, two and three respectively. For the Spearman correlation, Eculidean is 0.972, Manhattan is 0.982 and Minkowski is 0.955. Thus, Euclidean distance gives the best correlation results, while for alternatives, Manhattan distance or Minkoesky distance can be used.

Detection of maize seed germination rate based on improved locally linear embedding

Background . Transformer-based models have become central to natural language processing, demonstrating state-of-the-art performance in semantic similarity assessment, a task critical for various applications. These models capture detailed relationships between text, advancing the ability to gauge semantic relatedness. Materials and Methods. The performance of sentence embedding models, including all-mpnet-base-v2 , all-MiniLM-L6-v2 , paraphrase-multilingual-mpnet-base-v2 , bge-base-en-v1.5 , all-roberta-large-v1 , a ll-distilroberta-v1 , LaBSE , paraphrase-MiniLM-L3-v2 , bge-large-en-v1.5 , was assessed across different dataset sizes with two datasets. The following preprocessing steps were applied to the datasets: lowercasing, removing stop words, cleaning from special symbols and numbers, and lemmatization. Cosine similarity scores with negative values, indicating semantic dissimilarity, were treated as equivalent to a human-annotated similarity score of 0, and non-negative cosine similarity values were scaled to the 0-5 range. Metrics such as R 2 , MSE, RMSE, MAE, Spearman’s Correlation Coefficient, and Kendall's Tau were used for evaluation. Results and Discussion. Models’ performance generally improves with increased data. Evaluation of sentence embedding models revealed performance variations. all-roberta-large-v1 showed strong accuracy with high R 2 values and low errors. BAAI/bge-large-en-v1.5 excelled in capturing semantic relationships, demonstrating high Spearman’s and Kendall's Tau coefficients. all-MiniLM-L6-v2 demonstrated the fastest embedding generation. BAAI/bge-base-en-v1.5 presented the lowest accuracy. Processing times generally increase with data size. Conclusion. This study highlights a trade-off between accuracy and efficiency in sentence embedding. Model selection depends on balancing these factors to align with specific application needs. In cases when requiring high accuracy should favor all-roberta-large-v1 , while those prioritizing speed would benefit from all-MiniLM-L6-v2 . BAAI/bge-large-en-v1.5 is most suitable for tasks demanding semantic understanding of text details. Keywords : semantic similarity, sentence embeddings, transformers.

This paper presents a method to measure the semantic relatedness between sentences by using conceptual graph. To compare the sentence meaning, we firstly convert an English sentence into a conceptual graph. Then we employ the conceptual graph operations to match two conceptual graphs of one sentence and another. In matching nodes in the conceptual graphs, we utilize two lexicons i.e. WordNet and VerbNet. All semantic relatedness of contents in concept nodes are computed by using WordNet. The VerbNet is another used to find the semantic relatedness of contents in conceptual relation nodes. By our specified rules, the semantic relatedness between two sentences are objectively scored. We evaluate the performance of the proposed measurement with "Microsoft Research Paraphrase Corpus". The experimental results show the % correctness as 80.00% compared to human judgment. Moreover, we apply the measurement with words sense ambiguity analysis, the proposed measurement yields 76.44% of correctness compared to human judgment.

Accurately predicting building heating and cooling loads is crucial for optimizing HVAC systems and enhancing energy efficiency. However, data-driven models often face overfitting issues due to scarce training data, a common challenge for new constructions or under data privacy constraints. Transfer learning (TL) offers a solution, but its effectiveness heavily depends on selecting an appropriate source domain through effective similarity measurement. This study systematically evaluates the performance of 20 prevalent similarity metrics in TL for building heating load forecasting to identify the most robust metrics for mitigating data scarcity. Experiments were conducted on data from 500 buildings, with seven distinct low-data target scenarios established for a single target building. The Relative Error Gap (REG) was employed to assess the efficacy of transfer learning facilitated by each metric. The results demonstrate that distance-based metrics, particularly Euclidean, normalized Euclidean, and Manhattan distances, consistently yielded lower REG values and higher stability across scenarios. In contrast, probabilistic measures such as the Bhattacharyya coefficient and Bray–Curtis similarity exhibited poorer and less stable performance. This research provides a validated guideline for selecting similarity metrics in TL applications for building energy forecasting.

Introduction Large language models (LLMs) have the potential to realize accurate risk stratification and disease prediction by integrating multimodal data, such as electronic health records, medical images, and genomic profiles. However, in complex tasks like atherosclerosis risk estimation, the design of the prompt is critical to bring out the performance of LLMs. Several researches have demonstrated that in-context learning, where the prompt contains some cases, improves the performance of LLMs when the contained cases are carefully selected. Because manual selection usually requires enormous costs, automatic case selection has been desired. Methods This study aims to improve the effectiveness of in-context learning in LLM for atherosclerosis risk prediction by introducing a strategy for case selection. Our method involves two key components: feature selection and case selection. In the feature selection phase, we compute the mutual information between the clinical features and atherosclerosis risk using the database to extract the essential clinical features. This feature selection removes the unimportant features that disturb the performance of the case selection mechanism. Then, in the case selection phase, we select several cases from the database that are the most similar to the patient being diagnosed. This study compares three metrics to compute the similarity score: - Mahalanobis distance (trained via large margin nearest neighbor classification; LMNN [1]) - Cosine similarity (between the vector representations in KNN-augmented in-context example selection; KATE [2]) - Euclidean distance (between the prompt features) This combined strategy supplies the LLM with informative cases and thereby enhances the diagnostic performance. Results We used a fine-tuned Llama3-8B model for our target language and analyzed a dataset of 117,709 cases. Each case contained 96 features that were collected from annual health check-up data. From this dataset, 1,000 cases each were randomly selected for validation and test datasets. The remaining data were used as the database in our system. We considered a binary classification task and assigned the positive label when the Cardio-Ankle Vascular Index (CAVI) was greater than 8.0. The proportion of selected features was adjusted in 10% increments to maximize the F1 score on the validation dataset. The result shows that our proposed method, the strategic feature and case selections, significantly improved the F1 score for the test dataset by around 2% compared with zero-shot prompt and random selection (see the attached table). Among three similarity metrics, Euclidean distance yielded the highest F1 scores. We also observed that the feature selection phase with the adjusted number of selected features improved the F1 score in almost all settings. Conclusion Our strategy for feature and case selections improved the performance of in-context learning for LLM-based atherosclerosis risk prediction.

Evaluation Of Recommendation System On A Movie Dataset

Content Based Image Retrieval (CBIR) is a process in which for a given query image similar images will be retrieved based on the image content similarity. Image content refers to its visual features, which are mathematical representations of a digital image. The image retrieval task primarily depends on image feature extraction and similarity measurement between the feature vectors. The performance of CBIR process not only decided by the optimum features extracted from the image but also on the proper choice of the similarity/dissimilarity measures (distance metrics). As the image features broadly diversified in terms of color, texture and shape based, using the same distance metric may not work well for all of these features. In this paper, first we presented overview of geometric and statistical distance metrics used in CBIR along with the comparative analysis of these measures on color and texture features. Color features extracted by computing color histograms in HSV space and texture features by wavelet decompositions. Geometrical distances such as Manhattan, Chebyshev, Euclidean, statistical distance metrics such as Cosine Similarity, Chi-square, KullbackLeibler, Jeffrey and cumulative statistical distance metrics of Kolmogorov-Smirnov, Cramer von Mises and Earthmover's distances were analyzed for feature similarity. We gave certain conclusions on the performance of all these distance metrics in terms of Mean Average Precision (MAP) and Recall rates with color and texture features respectively.

While satellite-based remote sensing offers a promising avenue for large-scale LCC estimations, the accuracy of evaluations is often decreased by mixed pixels, attributable to distinct farming practices and diverse soil conditions. To overcome these challenges and to account for maize intercropping with soybeans at different growth stages combined with varying soil backgrounds, a hyperspectral database for maize was set up using a random linear mixed model applied to hyperspectral data recorded by an unmanned aerial vehicle (UAV). Four methods, namely, Euclidean distance, Minkowski distance, Manhattan distance, and Cosine similarity, were used to compare vegetation spectra from Sentinel-2A with the newly constructed database. In a next step, widely used vegetation indices such as NDVI, NAOC, and CAI were tested to find the optimum method for LCC retrieval, validated by field measurements. The results show that the NAOC had the strongest correlation with ground sampling information (R2 = 0.83, RMSE = 0.94 μg/cm2, and MAE = 0.67 μg/cm2). Additional field measurements sampled at other farming areas were applied to validate the method’s transferability and generalization. Here too, validation results showed a highly precise LCC estimation (R2 = 0.93, RMSE = 1.10 μg/cm2, and MAE = 1.09 μg/cm2), demonstrating that integrating UAV hyperspectral data with a random linear mixed model significantly improves satellite-based LCC retrievals.

Recently, securing Copyright has become a hot research topic due to rapidly advancing information technology. As a host cover, watermarking methods are used to conceal or embed sensitive information messages in such a manner that it was undetectable to a human observer in contemporary times. Digital media covers may often take any form, including audio, video, photos, even DNA data sequences. In this work, we present a new methodology for watermarking to hide secret data into 3-D objects. The technique of blind extraction based on reversing the steps of the data embedding process is used. The implemented technique uses the features of the 3-D object vertex' discrete cosine transform to embed a grayscale image with high capacity. The coefficient of vertex and the encrypted picture pixels are used in the watermarking procedure. Additionally, the extraction approach is fully blind and is dependent on the backward steps of the encoding procedure to get the hidden data. Correlation distance, Euclidean distance, Manhattan distance, and the Cosine distance are used to evaluate and test the performance of the proposed approach. The visibility and imperceptibility of the proposed method are assessed to show the efficiency of our work compared to previous corresponding methods.

<p>Phase congruency is an edge detector and measurement of the significant feature in the image. It is a robust method against contrast and illumination variation. In this paper, two novel techniques are introduced for developing a low-cost human identification system based on face recognition. Firstly, the valuable phase congruency features, the gradient-edges and their associated angles are utilized separately for classifying 130 subjects taken from three face databases with the motivation of eliminating the feature extraction phase. By doing this, the complexity can be significantly reduced. Secondly, the training process is modified when a new technique, called averaging-vectors is developed to accelerate the training process and minimizes the matching time to the lowest value. However, for more comparison and accurate evaluation, three competitive classifiers: Euclidean distance (ED), cosine distance (CD), and Manhattan distance (MD) are considered in this work. The system performance is very competitive and acceptable, where the experimental results show promising recognition rates with a reasonable matching time.</p>