Long-Text Abstractive Summarization using Transformer Models: A Systematic Review

Telemetry can inform many scientific and research questions if a context exists for integrating individual studies into the larger body of literature. Creating cumulative distributions of post-tagging evaluation metrics would allow individual researchers to relate their telemetry data to other studies. Widespread reporting of standard metrics is a precursor to the calculation of benchmarks for these distributions (e.g., mean, SD, 95% CI). Here we illustrate five types of standard post-tagging evaluation metrics using acoustically tagged Blue Catfish (Ictalurus furcatus) released into a Kansas reservoir. These metrics included: (1) percent of tagged fish detected overall, (2) percent of tagged fish detected daily using abacus plot data, (3) average number of (and percent of available) receiver sites visited, (4) date of last movement between receiver sites (and percent of tagged fish moving during that time period), and (5) number (and percent) of fish that egressed through exit gates. These metrics were calculated for one to three time periods: early ( 5 days early in the study. On average, tagged Blue Catfish visited 9 (50%) and 13 (72%) of 18 within-reservoir receivers early and at the end of the study, respectively. At the end of the study, 73% of all tagged fish were detected moving between receivers. Creating statistical benchmarks for individual metrics can provide useful reference points. In addition, combining multiple metrics can inform ecology and research design. Consequently, individual researchers and the field of telemetry research can benefit from widespread, detailed, and standard reporting of post-tagging detection metrics.

Multi-modal Medical Image Fusion (MMIF) is an advancing field at the intersection of medical imaging, data science, and clinical diagnostics. It aims to integrate complementary data from various imaging modalities, such as MRI, CT, and PET, into a single, diagnostically superior composite image. The limitations of unimodal imaging, such as low spatial resolution, insufficient contrast, or incomplete functional characterization, have catalyzed the development of MMIF techniques to enable enhanced visualization, precise diagnosis, and personalized therapeutic planning. This review provides a comprehensive synthesis of the MMIF landscape, categorizing methodologies into five principal domains such as spatial, frequency-based, sparse representation, deep learning, and hybrid approaches. Each technique is critically evaluated for its advantages, limitations, and applicability in clinical settings. Preprocessing, registration, fusion execution, and validation are covered in this review, along with levels of fusion pixel, feature, and decision. The study reviews prominent public databases, including TCIA, OASIS, ADNI, MIDAS, AANLIB, and DDSM, comparing their imaging modalities, disease coverage, file formats, and accessibility. The evaluation of MMIF techniques is systematically addressed, providing a framework for objective performance assessment. An experimental setup is implemented on two datasets to assess the comparative efficacy of selected MMIF techniques utilizing quantitative evaluation variables such as SSIM, entropy, spatial frequency, and mutual information. The results highlight the effectiveness of hybrid and deep learning-based approaches in maintaining both anatomical detail and functional consistency across modalities. The review explores MMIF’s real-world clinical applications, including image-guided neurosurgery, spinal planning, stereotactic radiosurgery, orthopedic implant design, and oncology diagnostics. It also provides insights into future directions, such as explainable AI, federated learning, and integration with robotic surgeries. MMIF offers immense potential yet has limitations like registration errors, computational burdens, generation of artifacts, loss of specific information, and a lack of standardized evaluation metrics. Essentially, the study provides an analytical basis for healthcare experts, scientists, and engineers aiming to develop clinically scalable MMIF systems, which will become indispensable tools for improving diagnostic accuracy, treatment planning, and patient outcomes in modern healthcare.

A survey and analysis on automatic image annotation

Segmentation of brain MRI using an altruistic Harris Hawks’ Optimization algorithm

BackgroundThe exponential growth in patient data collection by healthcare providers, governments, and private industries is yielding large and diverse datasets that offer new insights into critical medical questions. Leveraging extensive computational resources, Machine Learning and Artificial Intelligence are increasingly utilized to address health-related issues, such as predicting outcomes from Electronic Health Records and detecting patterns in multi-omics data. Despite the proliferation of medical devices based on Artificial Intelligence, data accessibility for research is limited due to privacy concerns. Efforts to de-identify data have met challenges in maintaining effectiveness, particularly with large datasets. As an alternative, synthetic data, that replicate main statistical properties of real patient data, are proposed. However, the lack of standardized evaluation metrics complicates the selection of appropriate synthetic data generation methods. Effective evaluation of synthetic data must consider resemblance, utility and privacy, tailored to specific applications. Despite available metrics, benchmarking efforts remain limited, necessitating further research in this area.ResultsWe present SynthRO (Synthetic data Rank and Order), a user-friendly tool for benchmarking health synthetic tabular data across various contexts. SynthRO offers accessible quality evaluation metrics and automated benchmarking, helping users determine the most suitable synthetic data models for specific use cases by prioritizing metrics and providing consistent quantitative scores. Our dashboard is divided into three main sections: (1) Loading Data section, where users can locally upload real and synthetic datasets; (2) Evaluation section, in which several quality assessments are performed by computing different metrics and measures; (3) Benchmarking section, where users can globally compare synthetic datasets based on quality evaluation.ConclusionsSynthetic data mitigate concerns about privacy and data accessibility, yet lacks standardized evaluation metrics. SynthRO provides an accessible dashboard helping users select suitable synthetic data models, and it also supports various use cases in healthcare, enhancing prognostic scores and enabling federated learning. SynthRO’s accessible GUI and modular structure facilitate effective data evaluation, promoting reliability and fairness. Future developments will include temporal data evaluation, further broadening its applicability.

Steganography involves embedding a secret message within another message, image, or video. This practice obscures the message and ensures that only the intended receiver can read it. Generally, the cover quality significantly impacts the overall effectiveness of the steganography system. Nevertheless, limited audio steganography-related articles on cover selection analysis have been observed. Therefore, this systematic literature review (SLR) presented a comprehensive analysis of the steganography-based cover selection, including their contexts and contents. An extensive search for relevant articles was initially conducted in this SLR. Specific inclusion and exclusion criteria were employed to ensure that only relevant articles were selected. A total of 44 articles were then selected and analyzed to identify the parameters and evaluation metrics used in cover selection. Consequently, the analysis determined four primary categories containing the cover selection techniques: image, audio, text, and video-based cover selections. Several factors influencing the cover selection were also identified based on the articles, such as parameters used in different mediums, standard evaluation metrics on each medium, and correlation between metrics performance and parameters. These findings indicated that the frequently utilized parameters were message and cover sizes. The imperceptibility [Peak-Signal-Noise-Ratio (PSNR) and Signal-Noise-Ratio (SNR)] was also the highest evaluation metric used in determining the effectiveness of the cover selection technique. Overall, this SLR served as a valuable resource for steganography researchers, bridging the gap between two research fields [machine learning (ML) and steganography]. A clear direction could be provided to future researchers through advanced analysis using this SLR to improve ML-related cover selection methods.

Introduction: In order to assess how well conventional and hybrid pitch detection techniques perform in speech processing applications, a comparative analysis of the two types of methods is conducted. Method: A proposed hybrid approach, Proposed PEF+CEP, is examined alongside five traditional algorithms, namely Normalized Correlation Function (NCF), Pitch Estimation Filter (PEF), Log- Harmonic Summation (LHS), Summation of Residual Harmonics (SRH) and Cepstrum Pitch Determination (CEP). The effectiveness is evaluated using performance metrics like accuracy, specificity, sensitivity, and Gross Pitch Error (GPE). Results and Discussion: Our findings show that the accuracy and specificity of the traditional methods are impressive; the accuracy and sensitivity of the suggested hybrid method surpass their performance, with an astounding 98.8% accuracy and 99.2% sensitivity. Conclusion: Furthermore, the Proposed PEF+CEP method is a promising solution for accurate and dependable pitch detection in speech processing applications because it strikes a strong balance between computational efficiency and robustness. These results open up new avenues for research in the field of speech processing and demonstrate the potential of hybrid approaches.

This paper proposes a stock price prediction model based on a combination method of recurrent neural networks (RNN) and attention mechanisms. The goal is to improve the forecasting accuracy of financial time series data. By applying the attention mechanism to the RNN architecture, the model can learn dynamic importance weights for different time steps. This allows better capture of dominant features in stock price volatility. The study utilizes real Apple Inc.'s stock prices between January 2024 and January 2025 to build models and validate the models. Normalizing and implementing train-test strategies were used to preprocess data in a bid to enhance stability within the models. Experiments utilize mean squared error (MSE), mean absolute error (MAE), and the coefficient of determination (R2) as evaluation metrics for the model. Comparative evaluation was performed against a set of mainstream popular models. Results show that the constructed RNN+Attention model outperforms traditional RNN, LSTM, MLP, and Transformer models both in prediction accuracy and fitting performance. In addition, visualizations demonstrate a strong correspondence between predicted values and actual values. The curve of training loss shows a clear downward trend, further supporting the effectiveness and convergence of the model. This method provides an efficient and realistic technical solution to stock price prediction using deep learning.

The standard evaluation metric of automatic speech recognition (ASR) is the word error rate (WER), which measures the dissimilarity between recognized word sequences and their ground truth. Many training algorithms designed to reduce sequence-level errors such as WER have been proposed for hidden Markov model (HMM)-based ASR, e.g., state-level minimum Bayes risk (sMBR). However, these approaches cannot be used directly for encoder-decoder model based end-to-end ASR, because the encoder-decoder model employs very different mechanisms from HMM-based approaches. In this paper, we propose a new method for optimizing the encoder-decoder model based on a sequence-level evaluation metric. Since the WER is not directly differentiable, we adopt a policy gradient objective function to train the encoder-decoder model, which enables us to minimize the expected WER of the model predictions. This training method employs the scoring of multiple hypotheses as in the decoding stage while usual cross entropy training uses only the ground truth. Therefore, we can expect it to improve the decoding results of the encoder-decoder model. We perform experiments using the Tedlium corpus to demonstrate the potential of our proposed method for improving the recognition performance of the encoder-decoder model.

Research in imbalanced domain learning has almost exclusively focused on solving classification tasks for accurate prediction of cases labelled with a rare class. Approaches for addressing such problems in regression tasks are still scarce due to two main factors. First, standard regression tasks assume each domain value as equally important. Second, standard evaluation metrics focus on assessing the performance of models on the most common values of data distributions. In this paper, we present an approach to tackle imbalanced regression tasks where the objective is to predict extreme (rare) values. We propose an approach to formalise such tasks and to optimise/evaluate predictive models, overcoming the factors mentioned and issues in related work. We present an automatic and non-parametric method to obtain relevance functions, building on the concept of relevance as the mapping of target values into non-uniform domain preferences. Then, we propose SERA, a new evaluation metric capable of assessing the effectiveness and of optimising models towards the prediction of extreme values while penalising severe model bias. An experimental study demonstrates how SERA provides valid and useful insights into the performance of models in imbalanced regression tasks.

Story visualization is an underexplored task that falls at the intersection of many important research directions in both computer vision and natural language processing. In this task, given a series of natural language captions which compose a story, an agent must generate a sequence of images that correspond to the captions. Prior work has introduced recurrent generative models which outperform text-to-image synthesis models on this task. However, there is room for improvement of generated images in terms of visual quality, coherence and relevance. We present a number of improvements to prior modeling approaches, including (1) the addition of a dual learning framework that utilizes video captioning to reinforce the semantic alignment between the story and generated images, (2) a copy-transform mechanism for sequentially-consistent story visualization, and (3) MART-based transformers to model complex interactions between frames. We present ablation studies to demonstrate the effect of each of these techniques on the generative power of the model for both individual images as well as the entire narrative. Furthermore, due to the complexity and generative nature of the task, standard evaluation metrics do not accurately reflect performance. Therefore, we also provide an exploration of evaluation metrics for the model, focused on aspects of the generated frames such as the presence/quality of generated characters, the relevance to captions, and the diversity of the generated images. We also present correlation experiments of our proposed automated metrics with human evaluations.

As a predominant renewable energy source, solar energy has gained a significant prominence due to its sustainable characteristics and environmental benefits. Precise forecasting of photovoltaic power generation constitutes a critical requirement for optimal energy management and the maintenance of stability in electrical grid operations. This work proposes a novel predictive modeling framework that implements long-short-term memory (LSTM) neural networks through the KNIME Analytics Platform to predict solar energy production utilizing high-resolution sensor data. The comprehensive data set incorporates time series measurements of key solar parameters, including solar irradiance, global horizontal irradiance (GHI) power density, plane-of-array (POA) energy yield, and module temperature, acquired from an industrialgrade solar monitoring infrastructure. Data preprocessing techniques and visualizations were used to improve model performance and interpretability. The LSTM model effectively captured temporal dependencies in the data, with evaluation metrics including mean absolute error (MAE), mean squared error (MSE), and root mean squared error (RMSE). The implemented LSTM architecture successfully learned complex temporal patterns inherent in the solar generation data, with rigorous performance validation conducted using standard evaluation metrics: mean absolute error (MAE), mean squared error (MSE) and root mean squared error (RMSE). The results demonstrate that a well-optimized LSTM can effectively model long-term solar patterns without the computational overhead of hybrid architectures while maintaining competitive accuracy, making it more scalable and deployable for real-world solar energy forecasting. This contribution addresses a fundamental research gap in the field by providing a simplified yet highly accurate forecasting solution, particularly beneficial for grid operators and renewable energy planners seeking efficient and reliable predictions.

Broad scale remote sensing promises to build forest inventories at unprecedented scales. A crucial step in this process is to associate sensor data into individual crowns. While dozens of crown detection algorithms have been proposed, their performance is typically not compared based on standard data or evaluation metrics. There is a need for a benchmark dataset to minimize differences in reported results as well as support evaluation of algorithms across a broad range of forest types. Combining RGB, LiDAR and hyperspectral sensor data from the USA National Ecological Observatory Network’s Airborne Observation Platform with multiple types of evaluation data, we created a benchmark dataset to assess crown detection and delineation methods for canopy trees covering dominant forest types in the United States. This benchmark dataset includes an R package to standardize evaluation metrics and simplify comparisons between methods. The benchmark dataset contains over 6,000 image-annotated crowns, 400 field-annotated crowns, and 3,000 canopy stem points from a wide range of forest types. In addition, we include over 10,000 training crowns for optional use. We discuss the different evaluation data sources and assess the accuracy of the image-annotated crowns by comparing annotations among multiple annotators as well as overlapping field-annotated crowns. We provide an example submission and score for an open-source algorithm that can serve as a baseline for future methods.

This study explores the use of Artificial Neural Networks (ANN) and Markov Chain models to forecast India’s black tea exports to major international markets. While the Markov Chain approach provided a simplified, state-based view of export behaviour, the ANN models were designed to capture continuous patterns and subtle market dynamics. Forecasts were generated for the next five years across ten key importing countries. Model performance was assessed using standard evaluation metrics- Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE) and R-squared (R²). The results clearly favoured the ANN model, which consistently delivered more accurate and reliable forecasts. Notably, the United Arab Emirates (UAE) emerged as the top growth market, with predicted export values from India reaching ₹1.23 lakhs by the fifth year. Russia and the USA also showed strong forecasted demand, with expected values of ₹64767 lakhs and ₹45238 lakhs, respectively. These insights offer practical value for exporters, traders and policymakers by highlighting priority markets and supporting more informed decision-making. Overall, this research reinforces the importance of intelligent forecasting systems in managing the complexity of international tea trade.

We are to present the specialized quality evaluation metrics by completely understanding the difference between the BPM, software focused on the process management by meeting the requirements for new enterprise environments, and other software products and reflecting the characteristics of process based-software based on the standard software quality evaluation metrics.