Search Results (9)

Recent advancements in natural language processing (NLP) have been significantly driven by the development of large language models (LLMs). Despite their impressive performance across various language tasks, these models still encounter challenges when processing tabular data. This study investigates the optimization of fine-tuning strategies for LLMs specifically in the context of tabular data processing. The focus is on the effects of decimal truncation, multi-dataset mixing, and the ordering of JSON key–value pairs on model performance. Experimental results indicate that decimal truncation reduces data noise, thereby enhancing the model’s learning efficiency. Additionally, multi-dataset mixing improves the model’s generalization and stability, while the random shuffling of key–value pair orders increases the model’s adaptability to changes in data structure. These findings underscore the significant impact of these strategies on model performance and robustness. The research provides novel insights into improving the practical effectiveness of LLMs and offers effective data processing methods for researchers in related fields. By thoroughly analyzing these strategies, this study aims to establish theoretical foundations and practical guidance for the future optimization of LLMs across a broader range of application scenarios. Full article

Runoff is greatly influenced by changes in climate conditions. Predicting runoff and analyzing its variations under future climates are crucial for ensuring water security, managing water resources effectively, and promoting sustainable development within the catchment area. As the key step in runoff modeling, the calibration of hydrological model parameters plays an important role in models’ performance. Identifying an efficient and reliable optimization algorithm and objective function continues to be a significant challenge in applying hydrological models. This study selected new algorithms, including the strategic random search (SRS) and sparrow search algorithm (SSA) used in hydrology, gold rush optimizer (GRO) and snow ablation optimizer (SAO) not used in hydrology, and classical algorithms, i.e., shuffling complex evolution (SCE-UA) and particle swarm optimization (PSO), to calibrate the two-parameter monthly water balance model (TWBM), abcd, and HYMOD model under the four objective functions of the Kling–Gupta efficiency (KGE) variant based on knowable moments (KMoments) and considering the high and low flows (HiLo) for monthly runoff simulation and future runoff prediction in Tunxi basin, China. Furthermore, the identified algorithm and objective function scenario with the best performance were applied for runoff prediction under climate change projections. The results show that the abcd model has the best performance, followed by the HYMOD and TWBM models, and the rank of model stability is abcd > TWBM > HYMOD with the change of algorithms, objective functions, and contributing calibration years in the history period. The KMoments based on KGE can play a positive role in the model calibration, while the effect of adding the HiLo is unstable. The SRS algorithm exhibits a faster, more stable, and more efficient search than the others in hydrological model calibration. The runoff obtained from the optimal model showed a decrease in the future monthly runoff compared to the reference period under all SSP scenarios. In addition, the distribution of monthly runoff changed, with the monthly maximum runoff changing from June to May. Decreases in the monthly simulated runoff mainly occurred from February to July (10.9–56.1%). These findings may be helpful for the determination of model parameter calibration strategies, thus improving the accuracy and efficiency of hydrological modeling for runoff prediction. Full article

Running machine learning algorithms for image classification locally on small, cheap, and low-power microcontroller units (MCUs) has advantages in terms of bandwidth, inference time, energy, reliability, and privacy for different applications. Therefore, TinyML focuses on deploying neural networks on MCUs with random access memory sizes between 2 KB and 512 KB and read-only memory storage capacities between 32 KB and 2 MB. Models designed for high-end devices are usually ported to MCUs using model scaling factors provided by the model architecture’s designers. However, our analysis shows that this naive approach of substantially scaling down convolutional neural networks (CNNs) for image classification using such default scaling factors results in suboptimal performance. Consequently, in this paper we present a systematic strategy for efficiently scaling down CNN model architectures to run on MCUs. Moreover, we present our CNN Analyzer, a dashboard-based tool for determining optimal CNN model architecture scaling factors for the downscaling strategy by gaining layer-wise insights into the model architecture scaling factors that drive model size, peak memory, and inference time. Using our strategy, we were able to introduce additional new model architecture scaling factors for MobileNet v1, MobileNet v2, MobileNet v3, and ShuffleNet v2 and to optimize these model architectures. Our best model variation outperforms the MobileNet v1 version provided in the MLPerf Tiny Benchmark on the Visual Wake Words image classification task, reducing the model size by 20.5% while increasing the accuracy by 4.0%. Full article

In the task of person re-identification (re-ID), capturing the long-range dependency of instances is crucial for accurate identification. The existing methods excel at extracting local features but often overlook the global information of instance images. To address this limitation, we propose a convolution-based counterfactual learning framework, called PSF-C-Net, to focus on global information rather than local detailed features. PSF-C-Net adopts a parameter-sharing dual-path structure to perform counterfactual operations in the prediction space. It takes both the actual instance image and a counterfactual instance image that disrupts the contextual relationship as the input. The counterfactual framework enables the interpretable modeling of global features without introducing additional parameters. Additionally, we propose a novel method for generating counterfactual instance images, which effectively constructs an explicit counterfactual space, to reliably implement counterfactual strategies. We have conducted extensive experiments to evaluate the performance of PSF-C-Net on the Market-1501 and Duke-MTMC-reID datasets. The results demonstrate that PSF-C-Net achieves state-of-the-art performance. Full article

Addressing the challenges in diagnosing and classifying self-care difficulties in exceptional children’s healthcare systems is crucial. The conventional diagnostic process, reliant on professional healthcare personnel, is time-consuming and costly. This study introduces an intelligent approach employing expert systems built on artificial intelligence technologies, specifically random forest, decision tree, support vector machine, and bagging classifier. The focus is on binary and multi-label SCADI datasets. To enhance model performance, we implemented resampling and data shuffling methods to tackle data imbalance and generalization issues, respectively. Additionally, a hyper framework feature selection strategy was applied, using mutual-information statistics and random forest recursive feature elimination (RF-RFE) based on a forward elimination method. Prediction performance and feature significance experiments, employing Shapley value explanation (SHAP), demonstrated the effectiveness of the proposed model. The framework achieved a remarkable overall accuracy of 99% for both datasets used with the fewest number of unique features reported in contemporary literature. The use of hyperparameter tuning for RF modeling further contributed to this significant improvement, suggesting its potential utility in diagnosing self-care issues within the medical industry. Full article

Arboviruses are a group of viruses that are transmitted by arthropods, such as mosquitoes, and cause significant morbidity and mortality worldwide. Currently, there are only a few options, with restricted use, for effective vaccines against these viruses. However, recent advances in arboviral vaccine development have shown promising innovations that have potential in preclinical and clinical studies. Insect-specific viruses have been explored as a novel vaccine platform that can induce cross-protective immunity against related arboviruses. Nanoparticle-based vaccines have also been developed to enhance the immunogenicity and stability of viral antigens. Additionally, vaccines against mosquito salivary proteins that can modulate the host immune response and interfere with arboviral transmission are being explored. Synonymous recoding, such as random codon shuffling, codon deoptimization, and codon-pair deoptimization, is being investigated as a strategy to attenuate the replication of arboviruses in vertebrate cells, reducing the risk of reverting to wild-type virulence. Finally, mRNA vaccines have been developed to rapidly generate and express viral antigens in the host cells, eliciting robust and durable immune responses. The challenges and opportunities for arboviral vaccine development are outlined, and future directions for research and innovation are discussed. Full article

Hyperspectral images (HSIs) are characterized by hundreds of spectral bands. The goal of HSI is to associate the pixel with a corresponding category label by analyzing subtle differences in the spectrum. Due to their excellent local context modeling capabilities, Convolutional Neural Network (CNN)-based methods are often adopted to complete the classification task. To verify whether the patch-data-based CNN methods depend on the homogeneity of patch data during the training process in HSI classification, we designed a random shuffling strategy to disrupt the data homogeneity of the patch data, which is randomly assigning the pixels from the original dataset to other positions to form a new dataset. Based on this random shuffling strategy, we propose a sub-branch to extract features on the reconstructed dataset and fuse the loss rates (RFL). The loss rate calculated by RFL in the new patch data is cross combined with the loss value calculated by another sub-branch in the original patch data. Moreover, we construct a new hyperspectral classification network based on the Siamese and Knowledge Distillation Network (SKDN) that can improve the classification accuracy on randomly shuffled data. In addition, RFL is introduced into the original model for hyperspectral classification tasks in the original dataset. The experimental results show that the improved model is also better than the original model, which indicates that RFL is effective and feasible. Experiments on four real-world datasets show that, as the proportion of randomly shuffling data increases, the latest patch-data-based CNN methods cannot extract more abundant local contextual information for HSI classification, while the proposed sub-branch RFL can alleviate this problem and improve the network’s recognition ability. Full article

Protein engineering—the process of developing useful or valuable proteins—has successfully created a wide range of proteins tailored to specific agricultural, industrial, and biomedical applications. Protein engineering may rely on rational techniques informed by structural models, phylogenic information, or computational methods or it may rely upon random techniques such as chemical mutation, DNA shuffling, error prone polymerase chain reaction (PCR), etc. The increasing capabilities of rational protein design coupled to the rapid production of large variant libraries have seriously challenged the capacity of traditional screening and selection techniques. Similarly, random approaches based on directed evolution, which relies on the Darwinian principles of mutation and selection to steer proteins toward desired traits, also requires the screening of very large libraries of mutants to be truly effective. For either rational or random approaches, the highest possible screening throughput facilitates efficient protein engineering strategies. In the last decade, high-throughput screening (HTS) for protein engineering has been leveraging the emerging technologies of droplet microfluidics. Droplet microfluidics, featuring controlled formation and manipulation of nano- to femtoliter droplets of one fluid phase in another, has presented a new paradigm for screening, providing increased throughput, reduced reagent volume, and scalability. We review here the recent droplet microfluidics-based HTS systems developed for protein engineering, particularly directed evolution. The current review can also serve as a tutorial guide for protein engineers and molecular biologists who need a droplet microfluidics-based HTS system for their specific applications but may not have prior knowledge about microfluidics. In the end, several challenges and opportunities are identified to motivate the continued innovation of microfluidics with implications for protein engineering. Full article

Calibration of conceptual rainfall–runoff models (CRRM) for water-resource assessment (WRA) is a complicated task that contributes to the reliability of results obtained from catchments. In recent decades, the application of automatic calibration techniques has been frequently used because of the increasing complexity of models and the considerable time savings gained at this phase. In this work, the traditional Rosenbrock (RNB) algorithm is combined with a random sampling method and the Latin hypercube (LH) to optimize a multi-start strategy and test the efficiency in the calibration of CRRMs. Three models (the French rural-engineering-with-four-daily-parameters (GR4J) model, the Swedish Hydrological Office Water-balance Department (HBV) model and the Sacramento Soil Moisture Accounting (SAC-SMA) model) are selected for WRA at nine headwaters in Spain in zones prone to long and severe droughts. To assess the results, the University of Arizona’s shuffled complex evolution (SCE-UA) algorithm was selected as a benchmark, because, until now, it has been one of the most robust techniques used to solve calibration problems with rainfall–runoff models. This comparison shows that the traditional algorithm can find optimal solutions at least as good as the SCE-UA algorithm. In fact, with the calibration of the SAC-SMA model, the results are significantly different: The RNB algorithm found better solutions than the SCE-UA for all basins. Finally, the combination created between the LH and RNB methods is detailed thoroughly, and a sensitivity analysis of its parameters is used to define the set of optimal values for its efficient performance. Full article