To address the issues of insufficient resolution and weak feature discriminability in traditional spectro-grams for time-frequency representation of speech signals,this paper proposes an optimal fractional-order spectro-gram generation method based on a quantum-inspired Newton-Raphson algorithm,aiming to enhance the perfor-mance of speech recognition and classification tasks.First,quantum encoding is employed to initialize the popula-tion in the Newton-Raphson algorithm.Quantum rotation gates guide individuals to converge toward the optimal solution,while quantum mutation and catastrophe mechanisms are introduced to maintain population diversity and prevent premature convergence.Furthermore,simulated annealing and Lévy flight strategies are incorporated to enhances the algorithm's global search capability.Next,the audio signal is preprocessed by windowing and framing.A fractional-order spectrogram is generated via the fractional Fourier transform,which is then compressed in scale using a Mel filter bank to obtain a fractional-order Mel spectrogram.Finally,an adjustable fractional-order para-meter α is introduced to expand the degrees of freedom in time-frequency representation.By minimizing informa-tion entropy as the objective function,hyperparameters including α,frame length,and frame shift are adaptively optimized using the quantum-inspired Newton-Raphson algorithm to obtain the optimal fractional-order spectrogram.Simulation experiment results on the CEC 2022 benchmark functions,the public emotional speech recognition data-set RAVDESS,the sound classification dataset UrbanSound8K,and a self-constructed singing vowel pronunciation dataset demonstrate that,the proposed quantum-inspired Newton-Raphson algorithm exhibits superior global optimi-zation capability and higher stability in solving high-dimensional complex problems compared with existing optimi-zation algorithms.The generated optimal fractional-order spectrograms effectively concentrate signal energy and enhance feature separability,significantly outperforming traditional speech feature extraction methods in terms of accuracy,recall,and F1 score for speech recognition.This method provides a novel approach for high-precision feature extraction of complex speech signals,effectively improves speech recognition performance,and offers good robustness and application prospects.
To address the lack of measurement devices and limited data indicators in current sit-to-stand(STS)ba-lance assessments,this study developed a novel ergonomic measurement device to capture multi-point vertical pres-sure signals from the hands,feet,and hips during the STS process.The system enables synchronized acquisition of time-series pressure data across key body regions.To represent a typical gradient of balance ability from normal to various levels of decline,85 participants were recruited,including young adults,community-dwelling older adults,and elderly patients from rehabilitation hospitals.Raw multi-channel pressure signals were collected during the STS movement and preprocessed using filtering,summation,segmentation,normalization,and phase division.55 candidate features were extracted based on time,force,and center-of-pressure dimensions.After univariate testing,collinearity elimination,and forward stepwise regression,three key features were obtained:rate of force develop-ment in the feet,pressure-time integral of the hands,and minimum foot pressure.An ordinal Logistic regression model was constructed using these features to classify balance ability levels,and its performance was evaluated via 10-times repeated 5-fold cross-validation.The results show that the model achieves an Akaike Information Crite-rion(AIC)value of 39.195 and a Pseudo R² of 0.843;that three variables,namely,foot force development rate,hand pressure-time integral,and minimum foot pressure,are all statistically significant(P<0.05);and that the ave-rage classification accuracy from repeated cross-validation is 91.06%,indicating good model fit and robust predic-tive performance.Moreover,regression analysis results reveal that higher foot force development rate and higher minimum foot pressure are associated with better balance,while a higher hand pressure-time integral indicates poorer balance ability.Using the sit-to-stand ergonomic measurement device designed in this study to collect time-series data and performing ordinal Logistic regression modeling can achieve relatively accurate classification of ba-lance ability,offering a potential new method for assessing balance ability in older adults.
The task of multimodal product summary generation aims to produce concise and accurate summary texts that highlight the core selling points of products by integrating both textual and visual information.However,existing methods still face two major challenges.First,traditional evaluation metrics such as ROUGE,which are based on lexical overlap,fail to precisely measure the completeness of a summary in expressing key product information.Second,mainstream supervised fine-tuning paradigms struggle to capture users'implicit preferences regarding the relative importance of different product information components,resulting in a misalignment between generated summaries and actual usage requirements.To address these issues,this paper proposes a component-based summarization evaluation metric,termed CSE(Component-based Summary Evaluation),which comprehensively evaluates the expression of key information from two perspectives:Component Hit Accuracy(CHA)and Component Quantity Ratio(CQR).Furthermore,a preference optimization-based multimodal product summary generation model,PAMPS,is designed.The model aligns its outputs with users'preferences for product component expression through four training stages:supervised fine-tuning,summary resampling,component evaluation-guided preference pair construction,and direct preference optimization.Experimental results on the large-scale Chinese e-commerce dataset CEPSUM show that PAMPS achieves significant improvements on ROUGE metrics.In particular,compared with Qwen2-VL-SFT,Qwen2-VL-DPO-ROUGE improves ROUGE-1,ROUGE-2,and ROUGE-L by 0.25,0.44,and 1.21 percentage points on average,respectively,demonstrating superior overall summary generation quality.Under the proposed CSE evaluation framework,the Qwen2-VL-DPO-CSE model demonstrates particularly prominent improvements in the CHA metric,with an average relative improvement of about 4%,indicating that component-oriented preference optimization can effectively enhance the model's ability to capture and express core product features.The comprehensive experimental results fully validate the effectiveness and practical applicability of the proposed method in improving the quality of multimodal product summary generation.
The structured and networked planning of urban low-altitude air routes is an effective approach to alle-viating the shortage of low-altitude airspace resources and ensuring operational safety.Considering the actual requirements of route deployment and traffic flow in urban low-altitude logistics scenarios,this paper proposes a low-altitude"trunk-feeder"route network planning method based on combined clustering,with the layout of takeoff and landing sites and actual operational traffic volumes as input conditions.Initially,the optimal orientation of trunk route was determined using principal component analysis(PCA)and numerical optimization,respectively.Meanwhile,the quality of the clustering results was evaluated using a Gaussian distribution model and a geometric similarity model,and four differentiated route network clustering schemes were constructed.Secondly,a comprehensive evaluation system encompassing the total route network length,construction costs,and route intersection points was established to assess the performance of the four clustering methods.Finally,the feasibility and effectiveness of the method were validated using Shaoxing City as a case study.The results show that,compared with the initial planning scheme,the total construction cost of the route network decreases by 21.39%and the number of route intersections decreases by 34.41%under the high-density scenario.Under the medium-density scenario,the total construction cost decreases by 36.44%and the number of route intersections decreases by 16.92%.The findings indicate that the proposed"principal component analysis-geometric similarity"combined clustering method adopted in this pa-per can significantly improve the safety and cost-effectiveness of low-altitude operations,and can provide theoretical foundations and technical support for the scientific planning of urban low-altitude traffic systems.
To screen for Limosilactobacillus reuteri strains with excellent safety profiles and high antimicrobial activity,this study systematically evaluated the biosafety,carbohydrate metabolic characteristics,and antimicrobial potential of Limosilactobacillus reuteri GZMV125,a strain previously isolated and preserved in our laboratory,by integrating whole-genome analysis with multidimensional experimental validation,thereby exploring the genetic basis underlying its functional properties.Genomic analysis revealed that the strain harbors one circular chromosome and one plasmid.No critical virulence factors were identified,while a type Ⅲ polyketide synthase gene cluster was detected.Safety assessments confirmed that the strain exhibits no hemolytic activity,and its antibiotic susceptibility profile meets the safety criteria for probiotic strains,indicating favorable biosafety characteristics.Antimicrobial assays demonstrated that its cell-free fermentation supernatant exhibits significant inhibitory effects against a broad range of foodborne pathogens,with the strongest inhibition observed against Listeria monocytogenes.Carbohydrate metabolism studies showed that the strain can efficiently utilize diverse carbon sources,including glucose and xylose,and that its antimicrobial activity is regulated by carbohydrate availability.Xylose was identified as the opti-mal carbon source for inducing the strongest antimicrobial activity,which may be associated with increased viable cell density and enhanced acid production capacity resulting from carbohydrate utilization.These findings estab-lished a potential association among"genomic metabolic potential-carbohydrate utilization phenotype-antimicrobial function."In conclusion,Limosilactobacillus reuteri GZMV125 exhibits excellent safety,broad-spectrum antimicro-bial capability,and strong adaptability in carbohydrate utilization,demonstrating considerable potential for develop-ment as a probiotic formulation or food fermentation starter culture.This research provides a scientific basis and theoretical support for its food-grade industrial application.
