High Performance Research Articles

High performance P300 spellers using GPT2 word prediction with cross-subject training

ABSTRACT Amyotrophic lateral sclerosis (ALS) severely impairs patients’ ability to communicate, often leading to a decline in their quality of life within a few years of diagnosis. The P300 speller brain–computer interface (BCI) offers an alternative communication method by interpreting a subject’s EEG response to flashing characters presented on a grid interface. This paper addresses the common speed limitations encountered in training efficient P300-based multi-subject classifiers by introducing innovative ‘across-subject’ classifiers. We leverage a combination of the second-generation Generative Pre-Trained Transformer (GPT2) and Dijkstra’s algorithm to optimize stimuli and suggest word completion choices based on subjects’ typing history. Additionally, we employ a multi-layered smoothing technique to accommodate out-of-vocabulary (OOV) words. Through extensive simulations employing random sampling of EEG data from multiple subjects, we demonstrate significant speed enhancements in typing passages containing rare and OOV words. These optimizations result in approximately 10 % improvement in character-level typing speed and up to 40 % improvement in multi-word prediction. We demonstrate that augmenting standard row/column highlighting techniques with layered word prediction yields close-to-optimal performance. Furthermore, we explore both ‘within-subject’ and ‘across-subject’ training techniques, showing that speed improvements are consistent across both approaches.

Is a Rural Entrepreneurial Ecosystem Conducive to the Improvement of Entrepreneurial Performance? Evidence from Typical Counties of Rural Entrepreneurship and Innovation in China

Rural entrepreneurship is an important means to solve the problem of “rural decline” and is also the focus of the “rural revitalization” strategy. The rural entrepreneurship ecosystem directly affects entrepreneurial performance. Based on the configurational perspective, using the fuzzy-set qualitative comparative analysis (fsQCA) method, taking 85 typical rural innovation and entrepreneurship demonstration counties in China as research samples, this study explores the impact path of the rural entrepreneurship ecosystem composed of multiple factors on entrepreneurial performance and the complex causal mechanisms behind it. The results show that market size, human capital, financial capital, infrastructure (both hardware and software), and government scale cannot individually constitute the necessary conditions for high or non-high rural entrepreneurial performance; there exist two pathways driving high entrepreneurial performance in rural areas: a market-driven financing-and-intelligence integration pathway and a government-supported infrastructure-assisted pathway. Under certain conditions, there is a substitution relationship between rural entrepreneurial market allocation and government intervention; there are two paths driving non-high rural entrepreneurial performance, which are summarized into market–financing suppression and market–government suppression according to the core driving factors. By systematically analyzing the impact of the rural entrepreneurial ecosystem on entrepreneurial performance, and explaining the intrinsic logic and path of high and non-high entrepreneurial performance based on the configurational perspective, this paper provides a decision-making reference for further enhancing the entrepreneurial performance in rural China and realizing rural revitalization.

Development of High Performance Light Weight Concrete through incorporation of Steel Fibers onto Polyurethane Foam Concrete

A lightweight, high-performance concrete was developed using steel Fiber-reinforcement onto polyurethane foam (PUF) enhanced lightweight concrete. Fiber-reinforcement enhances strength parameters of concrete. But, the interaction between polyurethane foam’s ability to reduce density and the reinforcing effects of steel Fibers in terms of mechanical strength, shrinkage, chloride penetration, and crack resistance has not yet been explored. To overcome this limitation, investigations were done in this study to find compressive strength, split tensile strength, flexural strength, shrinkage, impact resistance, and chloride ion penetration on PUF incorporated with steel Fibers. Microstructural analysis was performed using scanning electron microscopy (SEM) to examine Fiber-concrete bonding. The results revealed that mixes with 1.5% steel Fiber content exhibit good performance, in terms of improved compressive and flexural strengths, impact resistance, and shrinkage reduction. Chloride penetration test results exhibited good chloride resistance in Fiber-reinforced mixes, making them suitable for use in corrosive environments. SEM analysis confirmed that the improved mechanical properties were largely due to enhanced Fiber-matrix bonding, which contributed to better crack control. Thus, this study was able to develop a Polyurethane foam light weight concrete incorporated with steel Fibers which could synergistically enhance its strength and durability characteristics.

Decoding viewer emotions in video ads

Understanding and predicting viewers’ emotional responses to videos has emerged as a pivotal challenge due to its multifaceted applications in video indexing, summarization, personalized content recommendation, and effective advertisement design. A major roadblock in this domain has been the lack of expansive datasets with videos paired with viewer-reported emotional annotations. We address this challenge by employing a deep learning methodology trained on a dataset derived from the application of System1’s proprietary methodologies on over 30,000 real video advertisements, each annotated by an average of 75 viewers. This equates to over 2.3 million emotional annotations across eight distinct categories: anger, contempt, disgust, fear, happiness, sadness, surprise, and neutral, coupled with the temporal onset of these emotions. Leveraging 5-second video clips, our approach aims to capture pronounced emotional responses. Our convolutional neural network, which integrates both video and audio data, predicts salient 5-second emotional clips with an average balanced accuracy of 43.6%, and shows particularly high performance for detecting happiness (55.8%) and sadness (60.2%). When applied to full advertisements, our model achieves a strong average AUC of 75% in determining emotional undertones. To facilitate further research, our trained networks are freely available upon request for research purposes. This work not only overcomes previous data limitations but also provides an accurate deep learning solution for video emotion understanding.

High-entropy doping for high-voltage LiCoO2 with enhanced electrochemical performances

Lithium cobalt oxide (LiCoO2), as a pioneering layered oxide cathode material for lithium-ion batteries (LIBs), possesses exceptional theoretical specific capacity and cycling stability, positioning it as a leading candidate for commercial LIB applications. Boosting the upper-limit of charge voltage of LiCoO2 is a promising strategy to increase the capacity of batteries, vital for advancing energy storage solutions from portable electronics to electric vehicles. Nevertheless, hybrid redox reactions at high voltages speed up oxygen evolution, electrolyte decomposition, and irreversible phase transitions, ultimately causing rapid capacity fading.To address these challenges, we present a novel approach to enhance the structural stability and electrochemical performances of LiCoO2 cathode via La-Al-Mg-Y high-entropy doping. This strategy significantly improves the high-voltage cycling stability of the cathode material by retaining 72.5 % of its initial capacity after 500 cycles at 5 C under 4.6 V. Furthermore, the structural change caused by the phase transition between the O3 and H1-3 phases is remarkably suppressed. This advancement in LIB cathode technology paves the way for the development of high performance batteries and it is crucial for meeting the growing demand for sustainable energy storage.

Natural Rubber Latex Modified High Performance Concrete

Natural Rubber Latex Modified High Performance Concrete

Enhancing Efficiency and Stability of Inverted Perovskite Solar Cells through Solution‐Processed and Structurally Ordered Fullerene

AbstractThe electron transporting layer (ETL) used in high performance inverted perovskite solar cells (PSCs) is typically composed of C60, which requires time‐consuming and costly thermal evaporation deposition, posing a significant challenge for large‐scale production. To address this challenge, herein, we present a novel design of solution‐processible electron transporting material (ETM) by grafting a non‐fullerene acceptor fragment onto C60. The synthesized BTPC60 exhibits an exceptional solution processability and well‐organized molecular stacking pattern, enabling the formation of uniform and structurally ordered film with high electron mobility. When applied as ETL in inverted PSCs, BTPC60 not only exhibits excellent interfacial contact with the perovskite layer, resulting in enhanced electron extraction and transfer efficiency, but also effectively passivates the interfacial defects to suppress non‐radiative recombination. Resultant BTPC60‐based inverted PSCs deliver an impressive power conversion efficiency (PCE) of 25.3 % and retain almost 90 % of the initial values after aging at 85 °C for 1500 hours in N2. More encouragingly, the solution‐processed BTPC60 ETL demonstrates remarkable film thickness tolerance, and enables a high PCE up to 24.8 % with the ETL thickness of 200 nm. Our results highlight BTPC60 as a promising solution‐processed fullerene‐based ETM, opening an avenue for improving the scalability of efficient and stable inverted PSCs.

Nested Neural Networks: A Novel Approach to Flexible and Deep Learning Architectures

It’s easy to get lost in the complexities of sentences like, “Never will I have been in a place whose experiences will have been teaching me how amazing life really is till the beginning of every one of those marvelous whole new advenchallenges for more than 20 years of life by the end of my military service.” These kinds of nested and layered statements are not just the domain of advanced grammar—they have a parallel in cutting-edge technology as well. This paper introduces a novel neural network architecture, termed Nested Neural Networks (NNN), that incorporates nested layers within a more complex overarching structure. Much like the layered conditions in Future Perfect and Future Perfect Progressive tenses, NNNs utilize ‘nested’ conditions to bring flexibility and depth to data processing. This structure captures complex relationships, achieving high performance, computational efficiency, and generalization across tasks, much as understanding the nested clauses of a sentence can reveal deeper meanings. Our experiments on benchmark datasets demonstrate NNNs’ ability to handle intricate data patterns with enhanced memory efficiency and training adaptability, often outperforming traditional models. This fascinating intersection between language and machine learning showcases the powerful potential of structured complexity, whether in communication or computation.

The effect of the menstrual cycle phases on back squat performance, jumping ability and psychological state in women according to their level of performance -a randomized three-arm crossover study

ObjectiveThe influence of the menstrual cycle on practical power performance such as barbell back squats and jumping performance in women has not yet been fully investigated. In addition, the performance level of athletes has not been considered in previous studies. This study aimed to investigate the influence of different cycle phases on acute back squat performance, jumping ability and psychological state concerning the performance level.Methods24 female strength athletes (age: 25.2 ± 3.3 years; height: 169.5 ± 3.4 cm; body weight: 67.7 ± 7.3 kg) were recruited for the study. Level of performance was classified according to Santos et al. (intermittent (n = 13), advanced (n = 6), highly advanced (n = 5)). Participants were tested for 1RM barbell back squat and jumping performance (countermovement and squat jump) as well as two questionnaires assessing their psychological states in the menses (M), late follicular phase (FP) and mid-luteal phase (LP) in three MC. Saliva estradiol and progesterone concentrations with a menstrual cycle diary were used to confirm a normal MC. A principal components analysis for power performance, well-being, relaxation and alertness was carried out and a linear mixed model was used for statistical evaluation.ResultsNo significant differences were found between the MC phases in performance scores (p > 0.05), readiness (p > 0.05) and alertness (p > 0.05). However, a high correlation between MC phases, performance level and back squat performance was detected. Correlation analyses indicate that a higher performance level results in a higher variation depending on the MC of the squat performance. For well-being, a significantly lower score was detected in M than in FP and LP.ConclusionIn general the performance score of the lower body is not influenced by the MC. If strength performance and jumping ability are considered separately, there are indications that strength capability is influenced at a higher performance level. In addition, individual variance was also observed, so this should also be considered. However, further studies are needed to confirm this assumption due to the small sample sizes of the individual performance levels.Trial registrationGerman registry for clinical studies (DRKS00034816, Date: 08/01/2024).

IoT Intrusion Detection Model Using RNN Algorithm

An Intrusion Detection System (IDS) is the method that is processed in the network layer of an IoT system. Machine learning techniques have been applied in the IDS and observed the higher performance in identifying the intrusion and malware. The existing method involved in IDS tends to be ineffective due to the drawbacks of big data, centralization, and low privacy. The existing method is also inefficient in handling the streaming data of IoT systems. Most of the method in the IDS has low efficiency in intrusion detection to increase the efficiency of the detection. In this research, the FPA is proposed in IoT intrusion detection to increase the efficiency of the detection.

Factor analysis on the performance of elite male fencing athletes.

We analyzed how stress and heart rate variability affect the performance of male fencing athletes expected to compete in various international events representing Korea. This study included 72 fencing athletes who participated in training and competition from January 2020 to December 2023. The stress was analyzed using the Sports Competition Anxiety Test (SCAT). Heart rate variability (HRV) was measured using an autonomous neurodiagnostic device (SA-6000). Heart rate variability measurement items were time domain analysis (heart rate, standard deviation of NN interval, and root mean square of the successful differences) and frequency domain analysis (total power, low frequency, and low frequency/high-frequency ratio). All athletes were classified into two categories: victory (high performance) and defeat (low performance) groups in the recent national championships. In SCAT, there was no significant difference between high and low performance. After comparing HRV, it was found that there is a significant difference in heart rates (P=0.030) and low frequency/high-frequency ratio (P=0.006) between high and low performance. heart rates and low-frequency/high-frequency ratio during the high performance were observed to be higher than during low performance. In our study, we found that SCAT does not affect performance, while HRV has been shown to have an impact on performance. These findings can be valuable for coaches and athletes in developing strategies to improve performance results, particularly for elite male fencing athletes.

Preparation of high performance CoCuAl@SiC catalytic ceramic membrane for nonradical degradation of sulfamethoxazole

Catalytic ceramic membranes (CCM) featuring catalytic and separating functions, show huge potential in water treatment. Nevertheless, the inability of catalytic ceramic membrane with high catalytic performance to be produced by current fabrication method are the bottleneck preventing their applications. Herein, the high performance CoCuAl@SiC CCM were successfully prepared by coating Al gel interlayer, hydrothermal reaction and annealing in reductive atmosphere. Annealing in reductive atmosphere play key role in enhancing the catalytic property of CoCuAl@SiC CCM. 99.3 % sulfamethoxazole (10 mg·L−1) degraded by CoCuAl@SiC CCM annealing at 300 °C and 0.5 mM PMS in 1 min under a wide pH range of 3.5–11.0 and in the presence of anions. Quenching tests confirmed the dominant role of singlet oxygen (1O2) in SMX degradation. The abundant oxygen vacancies and redox cycles of Co3+/Co2+ and Co2+/Co0 on CoCuAl@SiC CCM are crucial for PMS activation. In addition, the SMX degradation pathways were found to comprise deamination, hydroxylation, and ring open according to the degradation intermediates of SMX. This study not only provides an approach for the preparation of catalytic membranes with excellent catalytic activity, but also sheds a novel insight into the indispensable role of oxygen vacancy.

A one-step integrated forming and curing process for smart thin-walled fiber metal laminate structures with self-sensing functions

This study proposes and analyzes a novel one-step integrated forming and curing (IFC) process for thin-walled fiber metal laminates (FMLs) structures embedded with fiber Bragg grating (FBG) sensors, and have achieved both high-performance properties and self-sensing functions in the formed structures. A prototype machine and testing setup have been developed to validate the process's feasibility by manufacturing high-performance FMLs flat and curvature parts with effective self-sensing capabilities for real-time manufacturing and in-service monitoring. Numerical models considering curing-induced deformation and heat transfer during manufacturing have also been developed to support the analysis and validation of the self-monitoring capabilities of the intelligent FMLs parts. The results reveal that with proper control of pressure (e.g., 0.6 MPa) and time during forming and curing, high tensile and impact performance of FMLs can be maintained with embedded FBG, with less than a 3% loss. Additionally, the IFC process can effectively lead to an apparent reduction of springback deformation in the formed FMLs (more than 80%). The validation of the self-sensing function during the manufacturing process has been achieved by comparing the strain monitoring results with finite element (FE) simulation results during curing, with a minimum discrepancy of 2.0%. For the in-service self-sensing function, comparison between FE analysis and surface-fixed strain gauges during the compression instability test confirmed the efficacy of FBG sensors, with a minimum discrepancy of 4.3%. The results show that the proposed novel IFC process enables the successful manufacture of smart thin-walled FMLs parts with high shape accuracy and mechanical properties in a single step and holds significant promise for manufacturing self-sensing smart structures in the aerospace and aviation industries.

Adaptive Multi-Model Predictive Control with Optimal Model Bank Formation: Consideration of Local Models Uncertainty and Stability

The Multiple Model Control (MMC) structure comprises three main components: the model bank, controller bank, and supervisor algorithm. Precise design of these components is crucial for achieving high control performance within the MMC framework, albeit this effort is not without its challenges. These challenges involve optimizing the model and controller banks ensuring system stability when dealing with uncertainties in the local models and enabling smooth switching between model-controller pairs. This paper addresses these challenges by presenting a comprehensive approach. Firstly, the optimal model bank is designed using an automatic clustering approach. Then, the design of the adaptive multi-model predictive controller bank and a supervisor algorithm capable of performing soft switching are discussed. The proposed control system exhibits the capability to ensure closed-loop system stability within each individual subspace, as well as during the transitioning between distinct subspaces. This stability is preserved even in the face of inherent uncertainties associated with the local models comprising the model bank. To evaluate and validate the performance of the proposed control system, it is applied to a satellite attitude control system. The results confirm the effectiveness and performance of the control system. The proposed control system holds promise for controlling highly nonlinear, complex, or switched systems, ensuring closed-loop system stability, and achieving high control performance.

Prediction of Drug Pathway-based Disease Classes using Multiple Properties of Drugs

Background: Drug repositioning now is an important research area in drug discovery as it can accelerate the procedures of discovering novel effects of existing drugs. However, it is challenging to screen out possible effects for given drugs. Designing computational methods are a quick and cheap way to complete this task. Most existing computational methods infer the relationships between drugs and diseases. The pathway-based disease classification reported in KEGG provides us a new way to investigate drug repositioning as such classification can be applied to drugs. A predicted class of a given drug suggests latent diseases it can treat. Objective: The purpose of this study is to set up efficient multi-label classifiers to predict the classes of drugs. Methods: We adopt three types of drug information to generate drug features, including drug pathway information, label information and drug network. For the first two types, drugs are first encoded into binary vectors, which are further processed by singular value decomposition. For the third type, the network embedding algorithm, Mashup, is employed to yield drug features. Above features are combined and fed into RAndom k-labELsets (RAKEL) to construct multi-label classifiers, where support vector machine is selected as the base classification algorithm. Results: The ten-fold cross-validation results show that the classifiers provide high performance with accuracy higher than 0.95 and absolute true higher than 0.92. The case study indicates the novel effects of three drugs, i.e., they may treat new diseases. Conclusion: The proposed classifiers have high performance and are superiority to the classifiers with other classic algorithms and drug information. Furthermore, they have the ability to discover new effects of drugs.

Optimal replicator dynamic controller for semi active control of long span cable stayed bridge with considering special variability of seismic excitations

Cable-stayed bridges are lifeline infrastructures. However, due to their design and structural characteristics, they are sensitive to the vibrations. Thus, vibration control in these structures is essential. Semi-active control systems have gained significant attention due to their high performance and adaptability. However, the implementation of these systems has always faced serious challenges particularly when it comes to developing appropriate control algorithms because semi active devices have complex and non-linear characteristics. Inspired by evolutionary game theory, the author utilizes the replicator dynamic controller concept to optimize the performance of MR dampers to improve the vibration reduction of cable-stayed bridges. Two key parameters in the proposed control methodologies using replicator dynamics are the total population (total available resources or the sum of actuator forces) and the growth rate. Instead of using the sensitivity analysis that has been done in previous studies for vibration reduction of highway bridges using a replicator controller, the author used an evolutionary algorithm named the nondominated sorting genetic algorithm (NSGA-II) to create an optimal replicator dynamic controller for more effective vibration reduction. The proposed methodology is evaluated through its numerical application to a second-generation benchmark example based on the Bill Emerson Memorial Bridge, which considers a more realistic behavior of the cable-stayed bridge by accounting for multiple support excitations. The study indicates that the optimal replicator dynamic controller improves the vibration reduction performance of MR dampers. Specifically, deck displacement is reduced by 46 % compared to passive control system and shear forces at the tower base are reduced by 33 % compared to active control systems for the El Centro earthquake. The results indicate that the proposed Replicator Dynamic Controller optimized through NSGA-II provides a robust and effective solution for improving seismic resilience of cable-stayed bridges.

SIPSC-Kac: Integrating swarm intelligence and protein spatial characteristics for enhanced lysine acetylation site identification

Elucidation of post-translational modifications (PTMs), such as lysine acetylation (Kac), is crucial for understanding protein function and regulation. Although traditional experimental methods for identifying Kac sites are accurate, they are time-consuming and costly, leading to incomplete acetylome mapping. Computational approaches, particularly those incorporating machine learning, offer a rapid alternative, but face challenges owing to dataset imbalance, limited feature space, and the need for more effective feature-selection algorithms. To address these challenges, we present SIPSC-Kac, a novel computational method that integrates swarm intelligence algorithms with protein spatial characteristics to enhance the prediction of Kac sites. We used the AlphaFold system for spatial feature extraction and employed swarm intelligence for optimal feature selection, outperforming existing methods in terms of accuracy and computational efficiency. SIPSC-Kac demonstrated superior performance across multiple bacterial species, which was validated by its high performance in evaluation metrics. Our web server provides researchers with a user-friendly platform for Kac site prediction, thereby contributing to the advancement of bioinformatics and proteomic research. The SIPSC-Kac code and web server are accessible, thereby promoting broad applications in the scientific community.

Facile synthesis and first principles calculations of Li-MoS2/rGO nanocomposite for high-performance supercapacitor applications

Two-dimensional (2D) nanostructured materials such as graphene oxide nanosheets and molybdenum disulfide (MoS2) are potential candidates for electrochemical energy storage applications. The layered structure of MoS2 makes it a promising active electrode material for supercapacitors. Lithium (Li) interacted with MoS2/reduced graphene oxide (Li-MoS2/rGO) nanocomposite has been synthesized in two steps using a sonochemical dispersion method, and a one-pot hydrothermal method resulting in few-layered MoS2/rGO nanosheets of 2D heterointerfaces. The Li-MoS2/rGO nanocomposite exhibits a high specific capacitance of 173.3 F g−1 at a current density of 1 A g−1. Furthermore, an asymmetric supercapacitor device fabricated with Li-MoS2/rGO achieves a good energy density of 10.6 Wh kg−1 at a power density of 686.3 W kg−1 and outstanding cycling stability with 81.2 % capacity retention over 10,000 cycles. The dispersion of 2D-MoS2 in water is due to the strong electrostatic interface. The structural and electronic properties are theoretically calculated for pristine and Li-interacted MoS2/rGO heterostructure using density functional theory. A comparison of the density of states plots of the two heterostructures and the difference charge density plot of Li-MoS2/rGO heterostructure clearly brings out the importance of the presence of Li. A value of 64 μF cm−2 for the quantum capacitance for pristine MoS2/graphene heterostructure increases to a value 99 μF cm−2 corroborating the experimental observations and we assign this change to the transfer of charge from the Li atom to the outer S atoms in the MoS2/graphene heterostructure. This comprehensive approach, combining experimental synthesis with computational modelling, significantly advances the development and optimization of high-performance supercapacitor materials in the energy storage field. The Li-MoS2/rGO nanocomposite has excellent electrochemical specific capacitance and long-term cyclic stability and has been effectively used for supercapacitor applications with high electrochemical performance.

Efficient sodium ion transport enhanced by anionophilic sites in polymer-supported plastic crystal electrolyte for high performance sodium metal anode

The design of sodium metal batteries (SMBs) is seen as an excellent solution to solve the request of next-generation energy storage technologies with high energy density, high safety, low cost, and environmental friendliness. However, the direct use of sodium metal as an anode produces the dendritic sodium, which harms the normal operation of battery. Inspired by solid-state lithium metal batteries, solid-state electrolyte (SSE) designed for SMBs effectively solves this shortcoming. Herein, a novel SSE with mask-based polymer membrane supporting plastic crystal electrolyte with ZIF-67 (MPPZ) is designed. The experimental results and density functional theory (DFT) calculations indicate that the addition of ZIF-67 in MPPZ electrolyte availably fixes TFSI− and achieves high Na+ transference number. Additionally, a protective layer of NaF is formed in situ on the surface of sodium metal to avoid side reactions between Na and MPPZ electrolyte while further guide the uniform deposition of Na. Accordingly, Na||Na symmetric battery with NaF-coated Na metal anode and MPPZ electrolyte has a stable deposition and stripping of Na for 3000 h. The assembled room temperature solid-state sodium-sulfur battery (RTSSSSB) exhibits a good cycle stability. This study provides an innovative thinking for the construction of high performance dendrite-free SMBs.

Hydroxylated SiO2-modified {0 0 1}-TiO2 nanosheets as a surface multifunctional photocatalyst for enhanced degradation of gaseous toluene

The relatively weak activity and serious deactivation of anatase TiO2 nanosheet catalysts with highly exposed {001} facets ({001}-TiO2) limits its wide application in photocatalytic degradation of gaseous pollutants. Herein, we propose a hybrid photocatalyst composed of {001}-TiO2 and SiO2 with numerous hydroxyl groups (denoted as {001}-TiO2/HO-SiO2). The catalyst showed excellent performance for gaseous toluene degradation under UV illumination, with 99.2 % removal efficiency and 204.7 μmol L−1 CO2 yield (about 85.9 % mineralization efficiency) within 120 min. It also demonstrated superior durability over five consecutive cycles of toluene photodegradation. The modification of HO–SiO2 on {001}-TiO2 facilitate the selective adsorption of toluene on the catalyst surface by H-bonding interaction, and make the major poisoning species, benzoic acid, easily desorb from the surface, reducing the catalyst inactivation. Most importantly, we found that the formed TiOSi bond promotes the separation of photogenerated electron–hole pairs, and provides abundant available electrons for the formation of reactive oxygen radicals and the left holes for the conversion of toluene to benzyl radicals or oxidation of H2O to hydroxyl radicals, which results in the improved activity. This study highlights the importance of the introduction of secondary components with special functions on TiO2, which is helpful for the design of photocatalysts with high performance and anti-deactivation ability for treatment of volatile organic compounds.