Search

Estimating similarity using multiple similarity measures or machine learning prediction models is a popular solution to the link prediction problem. The Relation Pattern Deep Learning Classification (RPDLC) technique is proposed in this study, and it is based on multiple neighbor-based similarity metrics and convolution neural networks. The RPDLC first calculates the characteristics for a pair of nodes using neighbor-based metrics and impact nodes. Second, the RPDLC creates a heat map using node characteristics to assess the similarity of the nodes’ connection patterns. Third, the RPDLC uses convolution neural network architecture to build a prediction model for missing relationship prediction. On three separate social network datasets, this method is compared to other state-of-the-art algorithms. On all three datasets, the suggested method achieves the greatest AUC, hovering around 99 percent. The use of convolution neural networks and features via relational patterns to create a prediction model are the paper’s primary contributions.

Emotional labor claims its significance as the key indicator both of the psychological health of contemporary employees, and the productivity of service-based businesses depending upon genuine emotional input of employees. By far, research on emotional labor of employees in an organizational context is still lacking. This study aims to explore the relationships among emotional labor, organizational support, career competences and career commitment to investigate how emotional labor interacts with the organizational context and affects the career management of the employee. Data were collected from a sample of 387 frontline employees working at two luxury hotel brands in China. Structural equation modeling (SEM) was utilized to estimate the relationships among the constructs. It is demonstrated by the findings that organizational support mediates positively on emotional labor, which exerts positive influences on career competences and career commitment. Sound handling of emotional labor, boosted by a supportive organizational environment, has been ascertained to positively predict long-term career paths of the employees at the company. This study provides insights into how the tourism and hospitality industry can optimize the functions of emotional labor for in enhancing service quality and customer satisfaction, as well as promoting the psychological well-being of the employees.

Additional file 4. Supplementary Table 2. Consequences of splice site mutations.

Additional file 5. Supplementary Table 3. Genotype and clinical presentation of patients with EB pruriginosa.

Additional file 3. Supplementary Table 1. Mutations and clinical subtypes of EB in Taiwan.

Higher-order topological superconductors and superfluids have triggered a great deal of interest in recent years. While Majorana corner or hinge states have been studied intensively, whether superconductors and superfluids, being topological or trivial, host higher-order topological Bogoliubov excitations remains elusive. In this work, we propose that Bogoliubov corner excitations can be driven from a trivial conventional $s$-wave superfluid through mirror-symmetric local potentials. The topological Bogoliubov excited modes originate from the nontrivial Bogoliubov excitation bands. These modes are protected by mirror symmetry and robust against mirror-symmetric perturbations as long as the Bogoliubov energy gap remains open. Our work provides new insight into higher-order topological excitation states in superfluids and superconductors., Comment: 7 pages,6 figures

This paper proposes an adaptive penalized weighted mean regression for outlier detection of high-dimensional data. In comparison to existing approaches based on the mean shift model, the proposed estimators demonstrate robustness against outliers present in both response variables and/or covariates. By utilizing the adaptive Huber loss function, the proposed method is effective in high-dimensional linear models characterized by heavy-tailed and heteroscedastic error distributions. The proposed framework enables simultaneous and collaborative estimation of regression parameters and outlier detection. Under regularity conditions, outlier detection consistency and oracle inequalities of robust estimates in high-dimensional settings are established. Additionally, theoretical robustness properties, such as the breakdown point and a smoothed limiting influence function, are ascertained. Extensive simulation studies and a breast cancer survival data are used to evaluate the numerical performance of the proposed method, demonstrating comparable or superior variable selection and outlier detection capabilities., Comment: 8 Tables, 4 figures

Additional file 6: sFigure 6. The correlation among DIP2B expression with immunostimulator related genes and immunoinhibitor related genes and chemokine receptor related genes in pan-cancer. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Additional file 7: sFigure 7. The correlation among DIP2B expression with immunostimulator related genes and immunoinhibitor related genes and chemokine receptor related genes in subtypes of BRCA. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Additional file 1. Supplementary Figure 1. Schematic outline of research methods.

Additional file 2. Supplementary Figure 2. Varying degrees of severity of EB pruriginosa, from relatively mild to severe.

Most reinforcement learning algorithms seek a single optimal strategy that solves a given task. However, it can often be valuable to learn a diverse set of solutions, for instance, to make an agent's interaction with users more engaging, or improve the robustness of a policy to an unexpected perturbance. We propose Diversity-Guided Policy Optimization (DGPO), an on-policy algorithm that discovers multiple strategies for solving a given task. Unlike prior work, it achieves this with a shared policy network trained over a single run. Specifically, we design an intrinsic reward based on an information-theoretic diversity objective. Our final objective alternately constraints on the diversity of the strategies and on the extrinsic reward. We solve the constrained optimization problem by casting it as a probabilistic inference task and use policy iteration to maximize the derived lower bound. Experimental results show that our method efficiently discovers diverse strategies in a wide variety of reinforcement learning tasks. Compared to baseline methods, DGPO achieves comparable rewards, while discovering more diverse strategies, and often with better sample efficiency.

We study the simplest quantum lattice spin model for the two-dimensional (2D) cubic ferromagnet by means of mean-field analysis and tensor network calculation. While both methods give rise to similar results in detecting related phases, the 2D infinite projected entangled-pair state (iPEPS) calculation provides more accurate values of transition points. Near the phase boundary, moreover, our iPEPS results indicate that it is more difficult to pin down the orientation of magnetic easy axes, and we interpret it as the easy-axis softening. This phenomenon implies an emergence of continuous $U(1)$ symmetry, which is indicated by the low-energy effective model and has been analytically shown by the field theory. Our model and study provide a concrete example for utilizing iPEPS near the critical region, showing that the emergent phenomenon living on the critical points can already be captured by iPEPS with a rather small bond dimension., Comment: 14 pages, 9 figures, 1 Table

Projection-free online learning, which eschews the projection operation via less expensive computations such as linear optimization (LO), has received much interest recently due to its efficiency in handling high-dimensional problems with complex constraints. However, previous studies assume that any queried gradient is revealed immediately, which may not hold in practice and limits their applications. To address this limitation, we generalize the online Frank-Wolfe (OFW) algorithm and the online smooth projection-free (OSPF) algorithm, which are state-of-the-art LO-based projection-free online algorithms for non-smooth and smooth functions respectively, into a delayed setting where queried gradients can be delayed by arbitrary rounds. Specifically, the main idea of our generalized OFW is to perform an update similar to the original OFW after receiving any delayed gradient, and play the latest decision for each round. Moreover, the essential change on OSPF is to replace the sum of queried gradients, which is originally utilized in each update, with the sum of available gradients. Despite their simplicities, our novel analysis shows that under a relatively large amount of delay, the generalized OFW and OSPF enjoy the same regret bound as OFW and OSPF in the non-delayed setting, respectively.

Long-Tailed Semi-Supervised Learning (LTSSL) aims to learn from class-imbalanced data where only a few samples are annotated. Existing solutions typically require substantial cost to solve complex optimization problems, or class-balanced undersampling which can result in information loss. In this paper, we present the TRAS (TRAnsfer and Share) to effectively utilize long-tailed semi-supervised data. TRAS transforms the imbalanced pseudo-label distribution of a traditional SSL model via a delicate function to enhance the supervisory signals for minority classes. It then transfers the distribution to a target model such that the minority class will receive significant attention. Interestingly, TRAS shows that more balanced pseudo-label distribution can substantially benefit minority-class training, instead of seeking to generate accurate pseudo-labels as in previous works. To simplify the approach, TRAS merges the training of the traditional SSL model and the target model into a single procedure by sharing the feature extractor, where both classifiers help improve the representation learning. According to extensive experiments, TRAS delivers much higher accuracy than state-of-the-art methods in the entire set of classes as well as minority classes., Comment: paper under review

Parking demand forecasting and behaviour analysis have received increasing attention in recent years because of their critical role in mitigating traffic congestion and understanding travel behaviours. However, previous studies usually only consider temporal dependence but ignore the spatial correlations among parking lots for parking prediction. This is mainly due to the lack of direct physical connections or observable interactions between them. Thus, how to quantify the spatial correlation remains a significant challenge. To bridge the gap, in this study, we propose a spatial-aware parking prediction framework, which includes two steps, i.e. spatial connection graph construction and spatio-temporal forecasting. A case study in Ningbo, China is conducted using parking data of over one million records before and during COVID-19. The results show that the approach is superior on parking occupancy forecasting than baseline methods, especially for the cases with high temporal irregularity such as during COVID-19. Our work has revealed the impact of the pandemic on parking behaviour and also accentuated the importance of modelling spatial dependence in parking behaviour forecasting, which can benefit future studies on epidemiology and human travel behaviours., Comment: DeepSpatial '21: 2nd ACM SIGKDD Workshop on Deep Learning for Spatiotemporal Data, Applications, and Systems (https://cs.emory.edu/~lzhao41/venues/DeepSpatial2021/)

Obtaining standardized crowdsourced benchmark of computational methods is a major issue in data science communities. Dedicated frameworks enabling fair benchmarking in a unified environment are yet to be developed. Here we introduce Codabench, an open-source, community-driven platform for benchmarking algorithms or software agents versus datasets or tasks. A public instance of Codabench (https://www.codabench.org/) is open to everyone, free of charge, and allows benchmark organizers to compare fairly submissions, under the same setting (software, hardware, data, algorithms), with custom protocols and data formats. Codabench has unique features facilitating the organization of benchmarks flexibly, easily and reproducibly, such as the possibility of re-using templates of benchmarks, and supplying compute resources on-demand. Codabench has been used internally and externally on various applications, receiving more than 130 users and 2500 submissions. As illustrative use cases, we introduce 4 diverse benchmarks covering Graph Machine Learning, Cancer Heterogeneity, Clinical Diagnosis and Reinforcement Learning.

Nonalcoholic fatty liver disease (NAFLD) enhances the growth and recurrence of colorectal cancer (CRC) liver metastasis. With the rising prevalence of NAFLD, a better understanding of the molecular mechanism underlying NAFLD-associated liver metastasis is crucial. Tumor-associated macrophages (TAMs) constitute a large portion of the tumor microenvironment that promotes tumor growth. NOD-like receptor C4 (NLRC4), a component of an inflammasome complex, plays a role in macrophage activation and interleukin (IL)-1β processing. We aimed to investigate whether NLRC4-mediated TAM polarization contributes to metastatic liver tumor growth in NAFLD. Wild-type and NLRC4-/- mice were fed low-fat or high-fat diet for 6 weeks followed by splenic injection of mouse CRC MC38 cells. The tumors were analyzed 2 weeks after CRC cell injection. High-fat diet-induced NAFLD significantly increased the number and size of CRC liver metastasis. TAMs and CD206-expressing M2 macrophages accumulated markedly in tumors in the presence of NAFLD. NAFLD up-regulated the expression of IL-1β, NLRC4, and M2 markers in tumors. In NAFLD, but not normal livers, deletion of NLRC4 decreased liver tumor growth accompanied by decreased M2 TAMs and IL-1β expression in tumors. Wild-type mice showed increased vascularity and vascular endothelial growth factor (VEGF) expression in tumors with NAFLD, but these were reduced in NLRC4-/- mice. When IL-1 signaling was blocked by recombinant IL-1 receptor antagonist, liver tumor formation and M2-type macrophages were reduced, suggesting that IL-1 signaling contributes to M2 polarization and tumor growth in NAFLD. Finally, we found that TAMs, but not liver macrophages, produced more IL-1β and VEGF following palmitate challenge. Conclusion: In NAFLD, NLRC4 contributes to M2 polarization, IL-1β, and VEGF production in TAMs, which promote metastatic liver tumor growth.

This thesis is concerned with the valorization of the cellulose pulps produced using the ionoSolv pretreatment process. Experiments using both fresh and recycled ionic liquid, specifically triethylammonium hydrogen sulfate [TEA][HSO4], were conducted. Characterization of the isolated pulps was used to demonstrate the chemical and morphological effects of the ionoSolv pretreatment process on cellulose. Recycled ionic liquid contains solutes that impact the mechanism for lignocellulose biomass fractionation affecting the optimal pretreatment conditions. Pulp yields were consistently lower (demonstrating better pretreatment extraction, the pulps have a higher glucan content) and saccharification results were better when using recycled ionic liquid, demonstrating that the presence of these solute improved lignin extraction. In addition, it is demonstrated that protic ionic liquid pretreatment will eventually generate humins and pseudolignin, which are undesirable by-products of carbohydrate conversion which reduce saccharification yields and pulp purity. Experiments were conducted by doping the ionic liquid with 13C labelled and unlabeled sugars and their derivatives to ascertain the role of these solutes in the formation or prevention of pseudo-lignin formation. The re-deposition of these compounds onto the pulp is directly impacted by the purity of the pulp surface. The use of recycled ionic liquid in combination with additives such as HMF demonstrated a moderating effect on the formation of pseudolignin and humic compounds by increasing their solubility in the ionic liquid, preventing their re-deposition onto the surface of the pulp during pretreatment. In addition, lignin recovered from the pretreatment suggests the incorporation of carbohydrates, with lignin produced using recycled ionic liquid having higher molecular weights. The pulps produced during the pretreatment were submitted to subsequent processing to demonstrate additional potential value streams. Pulps were bleached to produce cellulose fibers and nanocellulose crystals. Fiber analysis suggests the pulps are of sufficient quality for materials production and various other industrial applications. Open Access

Supplemental Material for Evaluating and characterizing urban vibrancy using spatial big data: Shanghai as a case study by Bo Huang, Yulun Zhou, Zhigang Li, Yimeng Song, Jixuan Cai and Wei Tu in Environment and Planning B: Urban Analytics and City Science

Too much sunlight exposure would cause heat stress for people during the hot summer, although a minimum amount of sunlight is required for humans., Comment: 21 pages, 7 figures

To deal with changing environments, a new performance measure -- adaptive regret, defined as the maximum static regret over any interval, was proposed in online learning. Under the setting of online convex optimization, several algorithms have been successfully developed to minimize the adaptive regret. However, existing algorithms lack universality in the sense that they can only handle one type of convex functions and need apriori knowledge of parameters. By contrast, there exist universal algorithms, such as MetaGrad, that attain optimal static regret for multiple types of convex functions simultaneously. Along this line of research, this paper presents the first universal algorithm for minimizing the adaptive regret of convex functions. Specifically, we borrow the idea of maintaining multiple learning rates in MetaGrad to handle the uncertainty of functions, and utilize the technique of sleeping experts to capture changing environments. In this way, our algorithm automatically adapts to the property of functions (convex, exponentially concave, or strongly convex), as well as the nature of environments (stationary or changing). As a by product, it also allows the type of functions to switch between rounds.

For the link prediction problem, it commonly estimates the similarity by different similarity metrics or machine learning prediction model. However, this paper proposes an algorithm, which is called Relation Pattern Deep Learning Classification (RPDLC) algorithm, based on two neighbor-based similarity metrics and convolution neural network. First, the RPDLC extracts the features for two nodes in a pair, which is calculated with neighbor-based metric and influence nodes. Second, the RPDLC combines the features of nodes to be a heat map for evaluating the similarity of the node’s relation pattern. Third, the RPDLC constructs the prediction model for predicting missing relationship by using convolution neural network architecture. In consequence, the contribution of this paper is purposed a novel approach for link prediction problem, which is used convolution neural network and features by relation pattern to construct a prediction model.

Chinese philosophy may be viewed as disciplined reflections on the insights of self-cultivation. Etienne Balazs asserted that all Chinese philosophy is social philosophy and that, even if Chinese thinkers dwell upon metaphysical speculation, they will sooner or later return to the practical issues of the world here and now. This concern for the concreteness of the life-world gives the impression that the social dimension of the human condition features so prominently in the Chinese world of thought that the idea of the group takes precedence over conceptions of the individual self. The anthropological studies that contrast the Chinese sense of shame with the Western sense of guilt further enhance the impression that external social approval, rather than internal psychological sanction, defines the moral fabric of Chinese society. The prevalent sociological literature on the mechanism of ‘saving face’ as a key to understanding Chinese interpersonal relationships also stresses the centrality of external conditioning in Chinese ethics. If we follow this line of thinking, it is easy to assume that Chinese philosophers are preoccupied with neither the transcendent referent nor the inner psyche. They are not particularly interested in questions of ultimate reality such as the creator, the origin of the cosmos or the existence of God. Nor are they engrossed in problems of the mind such as consciousness, self-identity or moral choice. Indeed, Chinese philosophy as social philosophy seems exclusively immersed in issues of correct behaviour, familial harmony, political order and world peace. Even strands of thought that emphasize the aesthetic experience of the self are all intimately bound up with the highly ritualized world of human-relatedness. Actually the spirit of spontaneity, as a liberation from social constraints, should be appreciated in terms of a conscious reflection on and critique of society and thus inherently sociological. However, this widely held opinion of Chinese philosophy is seriously flawed. While it offers a common-sense picture of where the strength of Chinese thought lies, it does not address the underlying reasons or the actual processes that define the main trajectory of the Chinese modes of thinking. Wing-tsit Chan suggests a more comprehensive characterization of Chinese philosophy as humanism: ‘not the humanism that denies or slights a Supreme Power, but one that professes the unity of man and Heaven’ (Chan 1963: 3). It is crucial to note that ‘humanism’ so conceived is diametrically opposed to secular humanism as a distinctive feature of the Enlightenment mentality of the modern West. Western humanism emerged as a thorough critique of spiritualism and a radical departure from naturalism, or a sense of affinity with nature; it was the result of secularization. Chinese humanism, on the other hand, tends to incorporate the spiritual and naturalist dimensions in a comprehensive and integrated vision of the nature and function of humanity in the cosmos. The advantage of characterizing Chinese philosophy as humanistic rather than sociological is to open the possibility of allowing aesthetic, religious and metaphysical as well as ethical, historical and political perspectives to shape the contours of the Chinese reflective mind. This synthetic approach better captures the spirit of Chinese thought because it was historical and social change, rather than speculation, which was instrumental in the outgrowth of humanism as a defining characteristic of Chinese philosophy.

The Zhongyong (Doctrine of the Mean) has traditionally been ascribed to Zisi, the grandson of Confucius and the indirect teacher of Mencius. Although this ascription has been challenged by modern critical scholarship since the turn of the twentieth century, recent archaeological finds indicate that the traditional view is not without textual base. If the Zhongyong actually predated the Mengzi, it seems that a significant portion of the Liji (Book of Rites), of which the Daxue (Great Learning) and Zhongyong are chapters, contains documents of the fifth century bc. This fact alone merits a fundamental restructuring of classical Confucian chronology and reinterpretation of the Mencian line of the Confucian tradition.

Originally a chapter in the Liji (Book of Rites), one of the Five Classics in the Confucian tradition, the Daxue (Great Learning) has for centuries attained the status of a canon, arguably the most influential foundational text in East Asian Confucian humanism. When the great neo-Confucian thinker Zhu Xi grouped the Daxue with the Zhongyong (Doctrine of the Mean), another chapter in the Liji, the Confucian Analects and the Mengzi as the Four Books, its prominence in the Confucian scriptural tradition was assured. Since the Four Books with Master Zhu’s commentaries became the required readings for the civil service examinations in 1313, and since Master Zhu insisted that the Daxue must be studied first among the Four, it has been widely acknowledged as the quintessential Confucian text.

Primary skin cancer can be divided into basal cell carcinoma(BCC), squamous cell carcinoma and melanoma three types. Especially, BCC incidence is as high as 80% with 10% global incidence increasing each year. Although BCC has low metastatic and lower mortality than the other two types, this malignant tumor has a very high incidence. Therefore places a great burden on global healthcare expenditures. Imiquimod and Resiquimod are therapies for BCC. Imiquimod has been shown to be highly effective in the treatment of localized BCC. Therefore, the purpose of this study is to analyze the microarray of therapeutic drugs for BCC using machine learning algorithm and constructing prediction model. In this study, microarray raw data with giving Imiquimod and Resiquimod to BCC, which contained 30,968 sets of gene expression. There are two parts of this study: First, the three gene selection algorithms were employed, such as linear regression analysis, information gain and gain ratio, to screen out the biomaker. Second, we use decision tree algorithms: C4.5 to construct the prediction model and discuss classification accuracy.

Machine learning techniques have deeply rooted in our everyday life. However, since it is knowledge- and labor-intensive to pursue good learning performance, human experts are heavily involved in every aspect of machine learning. In order to make machine learning techniques easier to apply and reduce the demand for experienced human experts, automated machine learning (AutoML) has emerged as a hot topic with both industrial and academic interest. In this paper, we provide an up to date survey on AutoML. First, we introduce and define the AutoML problem, with inspiration from both realms of automation and machine learning. Then, we propose a general AutoML framework that not only covers most existing approaches to date but also can guide the design for new methods. Subsequently, we categorize and review the existing works from two aspects, i.e., the problem setup and the employed techniques. Finally, we provide a detailed analysis of AutoML approaches and explain the reasons underneath their successful applications. We hope this survey can serve as not only an insightful guideline for AutoML beginners but also an inspiration for future research., Comment: This is a preliminary and will be kept updated

This chapter explores modernity as both a historical reality and a conceptual framework, and argues that Confucian East Asia helps to identify three sets of issues. These are traditions in the modernizing process, the relevance of non-Western civilizations to the self-understanding of the modern West, and the global significance of local knowledge. The Confucian insistence on the importance of equality rather than freedom, sympathy rather than rationality, civility rather than law, duty rather than rights, and human-relatedness rather than individualism may appear to be diametrically opposed to the value-orientation of the Enlightenment. It is unsurprising that the "Asian values" advocated by political leaders such as Lee Kwang Yew and Mahatir often provoke strong cynical reactions in the West. The Confucian tradition, marginalized as a distant echo of the feudal past, is forever severed from its imperial institutional base, but has yet kept its grounding in an agriculture-based economy, family-centered social structure, and paternalistic polity.

In this paper, seven wavelengths (465, 485, 508, 559, 588, 621 & 658nm) were generated synchronously by the single-pass potassium gadolinium tungstate (KGW) Raman laser, pumped by a 532nm picosecond Nd:YAG laser. To our knowledge, this Raman laser generated the highest order Stokes and anti-Stokes synchronously among all of the KGW Raman lasers by far. According to the experimental results, KGW crystal has the potential to become highly efficient solid-state Raman crystals to generate multi-wavelength laser.

Sephadex G-100 modified with chloracetic acid and CEAH was prepared into cationic and anionic carriers and actived by CNBr. β-galactosidase β-D-galactoside galacto hydrolase EC3. 2. 1. 23) isolated and partially purified from gram chicken bean was immobilized on modified Sephadex G-100 by means of adsorption of ions and crosslinked reaction. Both the anionic and cationic gel carriers have high protein binding capacity and high yield of enzyme activity. Kinetic results showed that the enzyme activity attained its maximum at 57°C for cationic carriers and 52°C for anionic carriers. In addition, the operational pH range of the immobilized preparations were increased. The immobilized enzymes showed good storage stability at room temperature than that of soluble enzyme. Results of repeated batch experiment suggested that the immobilized enzymes could be reused.

By four different methods, Acidithiobacillus ferrooxidans cells were immobilized by the complex of PVA and sodium alginate. The beads formed by these different methods were evaluated in terms of relative mechanical strength, biological activity, dilatability, and so on. The results indicate that the technique utilizing the complex of PVA and sodium alginate crosslinked with Ca(NO(3))(2) is more appropriate for the immobilization of A. ferrooxidans than any others. So the PVA-calcium nitrate beads were used in batch and continuous culture. A maximum ferrous iron oxidation rate of 4.6 g/l/h was achieved in batch culture. Long-time performance of packed-bed bioreactor was evaluated systematically over 40 days, depending on the conversion ratio of ferrous iron and the residence time. At a residence time of 2.5 h, 96% of the initial ferrous iron was oxidized. This study shows this new immobilization technique will be a feasible and economical method for A. ferrooxidans.

Seismic acquisition in the large desert area of the Junggar Basin in Western China faces three main challenges. First, three types of shielding effects (reflection, dissipation, and dispersion) make it difficult for seismic waves to propagate downwards. Second, near-surface interference waves and multiple interferences reduce the signal-to-noise (S/N) ratio of weak reflection waves from deep layers. Third, lateral variation of the very thick weathered zone makes it difficult to identify exploration targets. In response to these challenges, high-density, wide-azimuth geometry, a slipsweep with low-frequency vibrator, and a static correction based on shallow-layer reflection stacking were applied. In order to improve coupling of sources and receivers with the ground surface, seismic acquisition was carried out during the snowmelt season. The task was completed rapidly using a digital seismic system. New seismic data were significantly improved in comparison with previous datasets and strata previously hidden by shielding zones were clearly imaged.

A new Acidithiobacillus ferrooxidans cell immobilization technique utilizing the complex of PVA solution and sodium alginate solution crosslinked by Ca(NO 3 ) 2 as entrapment medium is reported. The mixture of A. ferrooxidans suspension and the entrapment complex were extruded into a solution of Ca(NO 3 ) 2 (1–5%) to form beads, then the beads were frozen at −20 °C for 1–2 days and thawed at room temperature. The forming mechanism, characteristic of this immobilized beads and the factors affecting activity of immobilized cells were also discussed. A maximum oxidation rate of 4.6 g Fe 2+ /(L h) was achieved in batch cultures by these immobilized cells. Precipitation formed during culture process was analyzed. The forming mechanism of this precipitation and how this precipitation affects the whole system were also discussed. In addition, the immobilization technique is operated simply, and the gel beads have high stability even under non-sterile conditions. So its application on an industrial scale would be more practicable.