Performance Advantages Research Articles

The Influence of Strategic Human Resource Practices and Ethical Leadership on Organizational Performance of Star Hotels: Mediated by Intellectual Capital and Sustainable Competitive Advantage

The study aims to analyze the influence of strategic human resource practices and ethical leadership on organizational performance in star-rated hotels in Jakarta, mediated by intellectual capital and sustainable competitive advantage. It addresses gaps in previous studies by examining both factors simultaneously and introducing decision making based on HR metrics. A quantitative method was employed, collecting data from HR supervisor level in star-rated hotels in Jakarta, with 255 respondents. Hypotheses were tested using PLS-SEM analysis. Results indicate that strategic human resource practices and ethical leadership positively impact organizational performance. Intellectual capital and sustainable competitive advantage mediate this effect. These findings underscore the importance of investing in intellectual capital and sustainable competitive advantage to enhance hotel performance. The study provides a reference for hotel management to develop effective HR and leadership strategies, emphasizing the importance of intellectual capital and competitive advantage. Regular monitoring and evaluation of these strategies are recommended to ensure alignment and positive impacts on hotel performance. Future research should consider adding variables such as technological readiness and innovation capability, given the rapid technological advancements and intense competition in the hospitality industry, which affect hotel performance and competitive advantage.

FINALLY: A Multi-Key FHE Scheme Based on NTRU and LWE

Multi-key fully homomorphic encryption (MKFHE), a generalization of fully homomorphic encryption (FHE), enables a computation over encrypted data under multiple keys. The first MKFHE schemes were based on the NTRU primitive, however these early NTRU based FHE schemes were found to be insecure due to the problem of over-stretched parameters. Recently, in the case of standard (non-multi key) FHE a secure version, called FINAL, of NTRU has been found. In this work we extend FINAL to an MKFHE scheme, this allows us to benefit from some of the performance advantages provided by NTRU based primitives. Thus, our scheme provides competitive performance against current state-of-the-art multi-key TFHE, in particular reducing the computational complexity from quadratic to linear in the number of keys.

Tunable crystallization behavior, memory effect, and thermo-mechanical properties of biobased polyamides PA5X

To reduce carbon dioxide emissions, the development of sustainable bio-based polyamides with performance advantages has become the focus of widespread attention. 1,5-Pentanediamine (PDA) is a renewable monomer produced by decarboxylation of lysine, which can be used as a sustainable alternative to traditional petroleum-based diamines and play a crucial role in the synthesis of bio-based polyamides. Herein, a series of bio-based aliphatic polyamides (PA5X) synthesized from PDA and aliphatic diacids with different methylene groups were successfully prepared via melt polycondensation. Their chemical structures were characterized, and the effects of diacid chain length on crystallization behavior, memory effect, and thermo-mechanical properties were emphatically investigated. The results show that the melting and crystallization temperatures of PA5X decrease with increasing diacid chain length. Meanwhile, the PA5X exhibits high thermal stability, with Td,5 % exceeding 380 °C. Unlike the even-even polyamide, the crystalline form of PA5X is γ-form, and the crystallization behavior and spherulitic morphology change significantly with diacid chain length, which is attributed to the complex entanglement effects. More particularly, the correlation between the diacid chain length and the melt memory effect is explored. PA56 with shorter methylene groups exhibits stronger melt memory effects due to the memory effects being directly affected by segmental chain interactions. Moreover, PA5X exhibits comparable or even superior tensile strength and ductility compared with hexamethylenediamine-based polyamide (PA6X) and reported fully bio-based PA11. This work provides a comprehensive investigation into the structure-property relationship of the bio-based polyamide PA5X, demonstrating great potential for application in high-performance eco-friendly materials.

SwinLabNet: Jujube Orchard Drivable Area Segmentation Based on Lightweight CNN-Transformer Architecture

Identifying drivable areas between orchard rows is crucial for intelligent agricultural equipment. However, challenges remain in this field’s accuracy, real-time performance, and generalization of deep learning models. This study proposed the SwinLabNet model in the context of jujube orchards, an innovative network model that utilized a lightweight CNN-transformer hybrid architecture. This approach optimized feature extraction and contextual information capture, effectively addressing long-range dependencies, global information acquisition, and detailed boundary processing. After training on the jujube orchard dataset, the SwinLabNet model demonstrated significant performance advantages: training accuracy reached 97.24%, the mean Intersection over Union (IoU) was 95.73%, and the recall rate was as high as 98.36%. Furthermore, the model performed exceptionally well on vegetable datasets, highlighting its generalization capability across different crop environments. This study successfully applied the SwinLabNet model in orchard environments, providing essential support for developing intelligent agricultural equipment, advancing the identification of drivable areas between rows, and laying a solid foundation for promoting and applying intelligent agrarian technologies.

Coordinated Ship Welding with Optimal Lazy Robot Ratio and Energy Consumption via Reinforcement Learning

Ship welding is a crucial part of ship building, requiring higher levels of robot coordination and working efficiency than ever before. To this end, this paper studies the coordinated ship-welding task, which involves multi-robot welding of multiple weld lines consisting of synchronous ones to be executed by a pair of robots and normal ones that can be executed by one robot. To evaluate working efficiency, the objectives of optimal lazy robot ratio and energy consumption were considered, which are tackled by the proposed dynamic Kuhn–Munkres-based model-free policy gradient (DKM-MFPG) reinforcement learning algorithm. In DKM-MFPG, a dynamic Kuhn–Munkres (DKM) dispatcher is designed based on weld line and co-welding robot position information obtained by the wireless sensors, such that robots always have dispatched weld lines in real-time and the lazy robot ratio is 0. Simultaneously, a model-free policy gradient (MFPG) based on reinforcement learning is designed to achieve the energy-optimal motion control for all robots. The optimal lazy robot ratio of the DKM dispatcher and the network convergence of MFPG are theoretically analyzed. Furthermore, the performance of DKM-MFPG is simulated with variant settings of welding scenarios and compared with baseline optimization methods. Compared to the four baselines, DKM-MFPG owns a slight performance advantage within 1% on energy consumption and reduces the average lazy robot ratio by 11.30%, 10.99%, 8.27%, and 10.39%.

Microstructure and mechanical properties of additively manufactured Ti–6Al–4V alloy based on large area, high-resolution EBSD mapping

The laser direct energy deposition (L-DED) is well-suited for metal manufacturing, offering advantages of excellent performance, high precision, intricate structures and material flexibility, etc. In this study, the microstructure and mechanical property of Ti–6Al–4V alloy were investigated using L-DED under various printing parameters. A large-area, high-resolution Electron Backscatter Diffraction (EBSD) mapping technique, enabling to characterize the coarse columnar β-grains with individual crystals up to a centimeter along the building direction, and the sub-grain acicular α′ phase down to 50 nm in the same image, was utilized to analyze the crystalline phases and their lattice orientations in additively manufactured Ti–6Al–4V parts. The microstructure evolution, involving the high temperature β grain structures, β+α phase transformation, as well as acicular α′ martensitic phases, were studied during the periodically attenuated thermal cycles of L-DED. The formation of the “fish-scale” morphology results primarily from the radically different α phase in size between the interior and edges of the “fish-scale” morphology. Additionally, the effect of processing parameters, i.e., scanning speed and laser power, on microstructures and mechanical properties is discussed in detail.

Noise analysis of variable-speed rotor technology on the coaxial lift-offset helicopter

The impact of variable-speed rotor technology on the aerodynamic and aeroacoustic characteristics of a coaxial lift-offset helicopter was investigated using a numerical analysis approach. The free wake vortex lattice method was utilized to predict aerodynamic performance, while the Ffowcs-Williams Hawkings acoustic analogy was employed to evaluate the noise impact. The Harrington coaxial rotor was used as the baseline configuration. Results show that reducing the rotational speed, in combination with lift-offset, leads to a decrease in overall power required during forward flight by lowering both collective and cyclic pitch. This is primarily due to a reduction in profile power across the speed range. However, excessive rotor deceleration can lead to an expanded reverse flow region, increasing induced power at high forward speeds. From an acoustics perspective, the reduced rotor speed mitigated loading noise radiated toward the ground, especially in the region 40-60 degrees below the rotor plane, achieving a maximum 2.6 dB reduction at the evaluated observer location. This study indicates that the technology can offer advantages for both performance and noise in forward flight conditions.

Adaptive robust integrated guidance and control for thrust-vector-controlled aircraft by solving LQR online

In order to enhance the matching relationship between guidance subsystem and control subsystem of the thrust-vector-controlled aircraft, a kind of sufficient modeling and adaptive robust design facing to integrated guidance and control (IGC) is proposed. With respect to the researched aircraft, a linear state-dependent mathematical model facing to IGC design is first established. Based on the as-built model, a new kind of adaptive robust IGC law is proposed and it is composed by an adaptive optimal IGC law and a robustness-improved IGC law. In order to guarantee the global stability of time-varying control system, the adaptive optimal IGC law is designed by solving Riccati matrix equation on line and using matrix Sign function method. Furthermore, in order to enhance the robust ability against the unmatched system uncertainties, the robustness-improved IGC law is designed by using dynamic surface control approach and extended state observing strategy. Simulation results present that, the proposed IGC scheme presents more performance advantages compared with traditional IGC schemes, including the improvement of guidance precision and attitude stabilization. Furthermore, in the conditions of 256 simulation combinations, the minimum relative distances between aircraft position and target position are distributed from 0.103m to 5.333m, and the average value is 2.208m, which means the proposed IGC scheme possesses strong robustness against different and time-varying model uncertainties.

Efficient Distance Queries on Non-point Data

Distance queries, including distance-range queries, k -nearest neighbors search, and distance joins, are very popular in spatial databases. However, they have been studied mainly for point data. Inspired by a recent approach on indexing non-point objects for rectangular range queries, we propose a secondary partitioning approach for space-partitioning indices, which is appropriate for distance queries. Our approach classifies the contents of each primary partition into 16 secondary partitions, taking into consideration the begin and end values of objects with respect to the spatial extent of the primary partition. Based on this, we define algorithms for three popular spatial query types, that avoid duplicate results and skip unnecessary computations. We compare our approach to the previous secondary partitioning method and to state-of-the-art data-partitioning indexing and find that it has a significant performance advantage.

Carbon matrix NiFeP/C nanoparticles anchored on Ti3C2Tx nanosheets as heterostructure high-performance anode for lithium-ion batteries

Engineered micro-nano heterogeneous architectures for transition metal phosphide (TMPs) anode materials in lithium-ion batteries (LIBs) provide multifunctional active components. These components synergistically interact to address pivotal challenges of pronounced volume expansion and sluggish reaction kinetics within the electrode. Consequently, this engineering design approach significantly enhances the overall battery performance. In this study, a self-assembled approach is employed to integrate hybridized NiFe-PBA nanocubes with two-dimensional (2D) monolayer MXene (Ti3C2Tx), producing a heterostructured NiFeP/C@Ti3C2Tx composite. In this meticulously engineered architecture, the interface between the outer MXene layer and NiFeP embedded within the carbon matrix exhibits several distinct superiorities, such as rapid interfacial ion migration, multidirectional migration pathways, and strong spatial adaptability. The resulting non-uniform phase structure effectively mitigates electrode expansion and enhances electronic conductivity. Consequently, the NiFeP/C@Ti3C2Tx-15 composite demonstrates distinct performance advantages over NiFeP/C without MXene. The composite achieves a specific capacity of 689.4 mAh g⁻¹ at 200 mA g⁻¹ over 300 cycles, double that of NiFeP/C. Furthermore, density functional theory (DFT) calculations validate the electronic conductivity facilitated by the heterointerface between MXene and NiFeP, substantiating the strategic significance of the heterostructure in advancing high-performance anode materials for lithium-ion batteries.

Unravelling the speed-going relationship: A proof of concept study from British turf flat and jump race meetings

The maximum galloping speeds of racehorses during a race are influenced by the functional performance of the ground (‘going’) amongst other factors. For turf racecourses in Britain, the ground is descriptively classified and numerically quantified on the morning of a race meeting by the clerk of the course and subsequently published to assist decision making. Importantly, this includes deciding whether a horse should or should not run. The going is also assessed and classified during the meeting by racing analysts using the normalized winning times from each race result. Differences between going assessments are regularly reported, therefore this study aimed to evaluate whether an alternative method of measuring going could better predict going measured from performances. Measurement and performance data from 25 flat and 25 jump meetings were compared using linear and nonlinear regression models. A continuous two-phase polynomial model for cushioning was found to be the best predictor of performance going for all 50 meetings (adjusted r2=0.819, P<0.001). As cushioning can provide a going related indicator of the forces that the horse will experience at gallop, this measurement may be useful when evaluating racecourse going. This initial model suggests that there is little performance advantage at maximal galloping speeds above a cushioning value of approximately10 kN, possibly due to changes in limb contact timings to manage limb forces limits as the ground becomes firmer. An expansion of objective measurements of going that relate to performance across a wider geographic region, if not internationally are needed to confirm this limit.

Fusing YOLOv5s-MediaPipe-HRV to Classify Engagement in E-learning: from the Perspective of External Observations and Internal Factors

The rapid advancements in computer vision technology present significant potential for the automatic recognition of learner engagement in E-learning. We conducted a two-stage experiment to assess learner engagement based on behavioural (external observations) and physiological (internal factors) cues. Using computer vision technology and wearable sensors, we extracted three feature sets: action, head posture and heart rate variability (HRV). Subsequently, we integrated our constructed YOLOv5s–MediaPipe behaviour detection model with a physiological detection model based on HRV to comprehensively evaluate learners’ behavioural, affective and cognitive engagement. Additionally, we developed a method and criteria for assessing distraction based on behaviour, ultimately creating a comprehensive, efficient, low-cost and easy-to-use system for the automatic recognition of learner engagement. Experimental results showed that our improved YOLOv5s model achieved a mean average precision of 92.2%, while halving both the number of parameters and model size. Unlike other deep learning-based methods, using MediaPipe–OpenCV for head posture analysis offers advantages in real-time performance, making it lightweight and easy to deploy. Our proposed long short-term memory classifier, based on sensitive HRV metrics and their normalisation, demonstrated satisfactory performance on the test set, with an accuracy = 80%, precision = 81%, recall = 80% and an F1 score = 80%.

Experimental study on seismic behavior of a novel RC slab- SC wall joint based on UHPC and rebar – steel plate lap splices

Steel plate concrete shear walls (SC walls) have been widely used in practical projects because of great mechanical properties. For the application in nuclear power plant containment, the joint of specially shaped reinforced concrete slab and SC walls is difficult to design and construct. By using ultra-high performance concrete (UHPC), a new type of joint with the non-contact lapping method is proposed in this paper, which has advantages of simple construction, superior performance, suitable cost, and wide application range. Five RC-UHPC slab-shear wall joints were designed and tested under reciprocating load to study the effects of structures of lapping steel plate, shear-resisting methods, and interfacial reinforcing methods on the mechanical properties of the joints. The test results showed that the joints adopting the new structures were damaged for the flexural failure of the concrete area, with no damage to the joint zone, while the other specimens failed in the shear failure of the joint zone. It can be demonstrated that the ribs formed by bars and studs on the lapping steel plates are effective measures for non-contact force transferring, and interfacial reinforcing bars can improve the strength of the interface between concrete and UHPC. The force transferring efficiency of the new joints was close to or more than 90%, which indicates that the proposed joint structures can effectively realize the non-contact lapping force transferring. The design formulas and suggestions are proposed based on the analysis of the mechanism of new structures.

How to combine different types of prefabricated components in a building to reduce construction costs and carbon emissions?

In the context of industrialized construction and carbon emission reduction globally, prefabrication has gained considerable attention. However, most studies cared the carbon emissions and construction costs from the perspective of a whole prefabricated building rather than the prefabricated components. It makes it difficult to understand the role of prefabricated components play and make improvements from the prefabricated components level. To fill this gap, this study proposes a model to optimize the selection and combination of prefabricated components in a building project to simultaneously reduce the construction costs and carbon emissions. Six common component types were considered: shear wall, beam, floor slab, stair, balcony, and air-conditioning slab. Their costs and carbon emissions were quantified when adopting prefabrication and cast-in-situ technologies. To obtain the optimal solution, we developed an improved multi-objective mayfly algorithm (IMOMA), which incorporates enhanced circle chaos mapping and a global best guiding strategy to generate Pareto solution sets. The performance of IMOMA was evaluated using the Zitzler-Deb-Thiele series test functions, demonstrating its advantages in convergence, diversity, and comprehensive performance through metrics such as generation distance, diversity metric Δ, and inverse generational distance. A case study validated the effectiveness of this proposed method, demonstrating an average reduction of 9.05% in total construction costs and 8.04% in carbon emissions when compared to the initial scheme. Further study revealed that different prefabrication rates correspond to different optimal components combination schemes, with different trends in the costs and carbon emission curves as the prefabrication rate increases. Additionally, carbon trading policies influence the components combination, with construction costs reduced by 10.88% at a carbon trading price of 42.80 CNY/t CO2eq. This study provides valuable guidance for general contractors to design optimal component combinations and for governments to promote the sustainable development of prefabricated construction.

Weakly Supervised Underwater Object Real-time Detection Based on High-resolution Attention Class Activation Mapping and Category Hierarchy

Recently, deep learning-based underwater object detection technology has achieved remarkable success. However, the accuracy and completeness of dataset instance annotation are crucial for its success. The quality of underwater images is low, severe objects clustering, and occlusion, acquiring object's annotations demands substantial time and labor costs, while mis annotation and missed annotation can also degrade model performance and limit their application in practical scenarios. To address this issue, this paper presents a novel weakly supervised underwater object real-time detection method, which is divided into two subtasks: weakly supervised object localization and real-time object detection. In the weakly supervised object localization task, we design a novel category hierarchy structure network that integrates the high-resolution attention-class activation mapping algorithm to obtain high-quality object class activation maps, weaken background interference, and obtain more complete object regions. The parameterized spatial loss module is devised to enable the model to escape from local optimal solutions, thus accurately and efficiently obtaining object pseudo-detection annotation boxes. For the real-time object detection task, the single-stage detector YOLOv7 is selected as the basic detection model, and an object perception loss function is designed based on the class activation map to jointly supervise the training process. A method for filtering noisy pseudo-supervision information is proposed to enhance the pseudo-supervision information involved in training. Ablation experiments and multi-method comparison experiments were conducted on the URPC and RUOD datasets, and the results verify the effectiveness of the proposed strategy, and our model exhibits significant advantages in detection performance and detection efficiency compared to current mainstream and advanced models.

A lightweight finger multimodal recognition model based on detail optimization and perceptual compensation embedding

Multimodal biometric recognition technology has attracted the attention of many scholars due to its higher security and stability than single-modal recognition, but its additional parameter quantity and computational cost have brought challenges to the lightweight deployment of the model. In order to meet the needs of a wider range of application scenarios, this paper proposes a lightweight model DPNet using fingerprint and finger vein images for multimodal recognition, which adopts a double-branch lightweight feature extraction structure combining detail optimization and perception compensation. Among them, the detail extraction optimization branch uses multi-scale dimensionality reduction filtering to obtain low-redundant detail information, and combines the depth extension operation to enhance the generalization ability of detail features. The perception compensation branch expands and compensates the model's perceptual field of view through lightweight spatial location query and global information attention. In addition, this paper designs a perceptual feature embedding method to embed perceptual compensation information in the way of importance adjustment to improve the consistency of embedded features. The ABFM fusion module is proposed to carry out multi-level lightweight and deep interactive fusion of the extracted finger modal features from the global to the spatial region, so as to improve the degree and utilization rate of feature fusion. In this paper, the model recognition performance and lightweight advantages are verified on three multimodal datasets. Experimental results show that the proposed model achieves the most advanced lightweight effect and recognition performance in the experimental comparison of all datasets.

A Patch Diversity Transformer for Domain Generalized Semantic Segmentation.

Domain generalization (DG) is one of the critical issues for deep learning in unknown domains. How to effectively represent domain-invariant context (DIC) is a difficult problem that DG needs to solve. Transformers have shown the potential to learn generalized features, since the powerful ability to learn global context. In this article, a novel method named patch diversity Transformer (PDTrans) is proposed to improve the DG for scene segmentation by learning global multidomain semantic relations. Specifically, patch photometric perturbation (PPP) is proposed to improve the representation of multidomain in the global context information, which helps the Transformer learn the relationship between multiple domains. Besides, patch statistics perturbation (PSP) is proposed to model the feature statistics of patches under different domain shifts, which enables the model to encode domain-invariant semantic features and improve generalization. PPP and PSP can help to diversify the source domain at the patch level and feature level. PDTrans learns context across diverse patches and takes advantage of self-attention to improve DG. Extensive experiments demonstrate the tremendous performance advantages of the PDTrans over state-of-the-art DG methods.

A Distributed Intrusion Detection Framework for Vehicular Ad Hoc Networks via Federated Learning and Blockchain

The emergence of connected vehicles via Vehicular Ad Hoc Networks (VANETs) has revolutionized transportation but has also brought forth challenges in security and privacy due to their open architecture. Early detection of intrusions within VANETs is paramount for ensuring safe communication. This research presents an intelligent distributed approach that leverages federated learning (FL) and blockchain for intrusion detection in VANETs. Through FL, various neural network models were implemented to distribute model training among vehicles, thus preserving privacy. Quantitative evaluation metrics demonstrate the effectiveness of the proposed framework. For example, compared to a traditionally trained Stochastic Gradient Descent (SGD) model, the Federated Trained Model achieved higher precision across various attack types, ranging from 68% to 94%, and consistently outperformed in terms of recall, with rates ranging from 57% to 88%. These results highlight FL's superiority in detecting intrusions, evidenced by gains in accuracy, recall, and precision. Integration of FL with blockchain further strengthened security and privacy protection, ensuring data integrity during collaborative FL training across decentralized nodes. This novel framework addresses VANET vulnerabilities by facilitating privacy-preserving, collaborative anomaly monitoring in a trustworthy manner. Evaluations validate the performance advantages of FL for intrusion identification, supporting wider adoption of vehicular technologies. The study underscores the potential of combining FL and blockchain to enable robust, cooperative abnormality recognition crucial for maintaining reliability, safety, and trust in VANET operations.

Optimizing kitchen ventilation with an integrated stove air supply-exhaust system for reducing PM2.5 intake fraction and enhancing energy efficiency

Ensuring good ventilation is crucial for reducing the pollution caused by cooking activities in the indoor environment. Among them, fume exhaust devices as a vital component of kitchen ventilation, and their performance is particularly critical. Merely increasing the air volume of exhaust devices to remove fumes not only leads to the higher energy consumption of exhaust fans, but also has limited practical effect on reducing pollution. For improving the ventilation condition and reducing the energy consumption of supply and exhaust fan, a new ventilation system for kitchen with integrated stove air supply-exhaust (ISASE) was developed in this study. Firstly, the orthogonal experiment was utilized for arranging different combination schemes of five influencing factors, including the emission rate, exhaust flow, upper air supply angle, upper and lower air supply velocities. Then, the effect of ISASE in reducing the intake fraction of PM2.5 under each scheme was studied by using the computational fluid dynamics method. The energy consumption of this system under different ventilation schemes was investigated with on-site testing. Finally, performance-gaining rate was proposed by introducing intake fraction reduction rate and energy consumption growth rate to quantitatively evaluate the performance advantage of ISASE. Polynomial fitting was also used to explore the energy-saving effect of ISASE at high emission rate. The results showed that at low, medium, and high emission rates, the intake fraction of PM2.5 was reduced by 65.0%–85.2%. However, the energy consumption of ISASE merely increased by 7.4%–16.8% compared with traditional integrated stove without air supply conditions. Its maximum performance-gaining rate reached 0.49–0.66. The energy-saving rate of ISASE was 46.7% compared with traditional integrated stove without air supply when the same intake fraction was achieved at high emission rate. The practical schemes of ISASE at different emission rates were given in this study, which optimized the working performance of integrated stove and provided a useful reference for its innovative design.

PocketDTA: an advanced multimodal architecture for enhanced prediction of drug-target affinity from 3D structural data of target binding pockets.

Accurately predicting the drug-target binding affinity (DTA) is crucial to drug discovery and repurposing. Although deep learning has been widely used in this field, it still faces challenges with insufficient generalization performance, inadequate use of 3D information, and poor interpretability. To alleviate these problems, we developed the PocketDTA model. This model enhances the generalization performance by pre-trained models ESM-2 and GraphMVP. It ingeniously handles the first 3 (top-3) target binding pockets and drug 3D information through customized GVP-GNN Layers and GraphMVP-Decoder. In addition, it uses a bilinear attention network to enhance interpretability. Comparative analysis with state-of-the-art (SOTA) methods on the optimized Davis and KIBA datasets reveals that the PocketDTA model exhibits significant performance advantages. Further, ablation studies confirm the effectiveness of the model components, whereas cold-start experiments illustrate its robust generalization capabilities. In particular, the PocketDTA model has shown significant advantages in identifying key drug functional groups and amino acid residues via molecular docking and literature validation, highlighting its strong potential for interpretability. Code and data are available at: https://github.com/zhaolongNCU/PocketDTA.