Search

This paper presents an improved teaching learning-based whale optimization algorithm (TSWOA) used the simplex method. First of all, the combination of WOA algorithm and teaching learning-based algorithm not only achieves a better balance between exploration and exploitation of WOA, but also makes whales have self-learning ability from the biological background, and greatly enriches the theory of the original WOA algorithm. Secondly, the WOA algorithm adds the simplex method to optimize the current worst unit, averting the agents to search at the boundary, and increasing the convergence accuracy and speed of the algorithm. To evaluate the performance of the improved algorithm, the TSWOA algorithm is employed to train the multi-layer perceptron (MLP) neural network. It is a difficult thing to propose a well-pleasing and valid algorithm to optimize the multi-layer perceptron neural network. Fifteen different data sets were selected from the UCI machine learning knowledge and the statistical results were compared with GOA, GSO, SSO, FPA, GA and WOA, severally. The statistical results display that better performance of TSWOA compared to WOA and several well-established algorithms for training multi-layer perceptron neural networks.

This study was conducted to determine the optimal dose of the oral solution of the ultrasonic extract of Radix dichroae (UERD) and to provide experimental support for a safe clinical dose for anticoccidial treatment of broiler chickens. Radix dichroae root extracts were prepared using the ultrasonic extraction method. The anticoccidial activity of the oral solution prepared from the ultrasonic extract of Radix dichroae roots was tested in broiler chickens following oral infection with a field isolate of E. tenella. Ninety Lingnan yellow broiler chickens (14 days old) were randomly divided into nine groups (n = 10), including six UERD oral solution treatments (0.25, 0.50, 1.50, 2.50, 3.50, and 5.00%), a toltrazuril group (0.10%), an E. tenella-infected control group, and a healthy control group. All groups were inoculated orally with 7 × 104 sporulated E. tenella oocysts (Guangdong strain) except for the healthy control group. The chickens in the seven drug-treated groups were administered a UERD oral solution or toltrazuril in drinking water for 7 days. The anticoccidial efficacy of the UERD oral solution was evaluated by the bloody diarrhoea severity level, relative body weight gain (rBWG), lesion score, oocyst per gram (OPG), and anticoccidial index (ACI). Compared with the infected control group, there were no significant differences in the groups treated with UERD oral solution or toltrazuril with regard to the lesion changes in the caecal regions (P>0.05); however, the blood contents, OPG, and oocyst score in three UERD oral solution treatment groups (0.50, 1.50, and 2.50%) were significantly reduced, and the bloody diarrhoea was also alleviated. The ACI in three UERD oral solution treatment groups (0.50%, ACI = 143.7; 1.50%, ACI = 151.0; and 2.50%, ACI = 144.3) was higher than that in the toltrazuril group (ACI = 127.0), and the rBWG in the 1.50% UERD oral solution treatment group (95.0%) was similar to that in the healthy control group (100%), which was also 12.5% higher than that in the toltrazuril group (82.5%). The findings of this study demonstrated that the UERD oral solution (0.50% ～ 2.50% dose range) showed better prevention, anticoccidial efficacy, and growth promotion effects than toltrazuril (0.10%), and the 1.50% dose level of UERD oral solution in water is the clinically recommended dose according to the present study conditions.

Semantic attributes have been proved to be an effective representation approach for image classification in recent years. However, most of the existing semantic attributes are explicitly pre-defined and the attribute number is also very limited in practice. Therefore, we often encounter difficulties by using a fixed incomplete semantic attribute set for different classification tasks. One possible solution is to expand the semantic attribute representation with some non-semantic mid-level features. However, how to make the expanded features more effective and discriminative is still seldom exploited. In this paper, we propose a Sequential augmented Attributes learning (SAL) method to implement semantic attribute augmentation. In our SAL method, the non-semantic mid-level features are learned one by one under a sequential error-correcting scheme so that we can obtain more discriminating power with very compact expanded features. Extensive experiments are conducted on two public datasets and the results show that our approach achieves encouraging performance.

Relevance feedback is an effective approach to improve the performance of image retrieval by leveraging the labeling of human. In order to alleviate the burden of labeling, active learning method has been introduced to select the most informative samples for labeling. In this paper, we present a novel batch mode active learning scheme for informative sample selection. Inspired by the method of graph propagation, we not only take the correlation between labeled samples and unlabeled samples, but the correlation among unlabeled samples taken into account as well. Especially, considering the unbalanced distribution of samples and the personalized feedback of human we propose an asymmetric propagation scheme to unify the various criteria including uncertainty, diversity and density into batch mode active learning in relevance feedback. Extensive experiments on publicly available datasets show that the proposed method is promising.

Recommendation for new users is a key challenge due to the lack of prior information from them, which is the well-known cold-start problem. Preference elicitation has been proposed as an efficient strategy for eliciting new users preference through an initial interview where new users are queried by elaborately selected items. In this paper, we propose a novel community discovering guided discriminative selection (CDDS) model for constructing query set. We exploit the community as an effective information which is not fully used in existing approaches. By integrating item selection and community discovery into one framework, our model selects most discriminative items for preference elicitation, with guidance of unsupervised community discovering process. To perform community discovering process, the model utilizes rating similarity graph and social network as a graph regular-ization. Experimental results on real-world datasets Flixster and Douban demonstrate that the proposed method outperforms traditional preference elicitation methods for cold-start recommendation.

Augmenting the semantic attribute representation with the discriminative features has been proved to be an effective method for improving the performance of object classification. However, how to make the expanded features more effective and discriminative is still an open problem. In this paper, we propose a Sequential Augmented features Learning method (SAL) to implement semantic attribute augmentation. In our SAL method, the augmented non-semantic features are learned one by one under a sequential error-correcting scheme so that we can obtain more discriminating power with very compact expanded features. Extensive experiments are conducted on a public dataset and the results show that our approach achieves encouraging performance.

The goal of relevance feedback is to improve the performance of image retrieval by leveraging the labeling of human. It is helpful to introduce active learning method in relevance feedback to alleviate the human burden. In the traditional active learning the samples which can improve the classifier the most if they were labeled are selected for the user's labeling. However, the change of the geometrical structure of the data distribution caused by such expensive labeled samples is not fully exploited. By mining user's labeling information, we can reduce the original feature space dimension to ease the classifier's instability brought by the small sample size. In this paper, we propose a novel batch mode active learning method for informative data selection. The labeled samples are not only used to retrain the classifier, but to learn a subspace which efficiently encodes user's intention as well. Especially, a scheme of certainty propagation on the subspace effectively integrates uncertainty sampling and subspace learning into the proposed Subspace learning based batch mode Active Learning method (SubAL) in relevance feedback. Extensive experiments on publicly available dataset shows that the proposed method is promising.

Traditionally, Collaborative Filtering assumes that similar users have similar responses to similar items. However, human activities exhibit heterogenous features across multiple domains such that users own similar tastes in one domain may behave quite differently in other domains. Moreover, highly sparse data presents crucial challenge in preference prediction. Intuitively, if users' interested domains are captured first, the recommender system is more likely to provide the enjoyed items while filter out those uninterested ones. Therefore, it is necessary to learn preference profiles from the correlated domains instead of the entire user-item matrix. In this paper, we propose a unified framework, TopRec, which detects topical communities to construct interpretable domains for domain-specific collaborative filtering. In order to mine communities as well as the corresponding topics, a semi-supervised probabilistic topic model is utilized by integrating user guidance with social network. Experimental results on real-world data from Epinions and Ciao demonstrate the effectiveness of the proposed framework.

Recommendation for cold users is fairly challenging because no prior rating can be used in preference prediction. To tackle this cold-start scenario, rating elicitation is usually employed through an initial interview in which users are queried by some carefully selected items. In this paper, we propose a novel framework to mine the most valuable items to construct query set using a semi-supervised discriminative selection (SSDS) model. To learn a low dimensional representation for users in item space which can reflect their tastes to a large extent, the model incorporates category labels as discriminative information. To ensure the used labels reliable as well as all users considered, the model utilizes a semi-supervised scheme leveraging expert guidance with graph regularization. Experimental results on real-world dataset MovieLens demonstrate that the proposed SSDS model outperforms traditional preference elicitation methods on top-N measures for cold-start recommendation.

How to align social tags with image regions without additional human intervention is a challenging but a valuable task since it can provide more detailed image semantic information and improve the accuracy of image retrieval. To this end, we propose a novel tag-to-region method with two phases of sparse reconstructions by exploring the large-scale user contributed resources. Given an image with social tags, we first explore the tagging information of large-scale social images to sparsely reconstruct the label vector of the given image, and then use the reconstructing weights as the semantic relevance to the image. With the top $T$ semantically relevant images, we further employ a group sparse coding algorithm to reconstruct each region of the given image, in which the regions from the social images with a common label are deemed as a label group. The group sparsity works on the assumption that one image region corresponds to tags as few as possible. Finally, the region-level tags can be predicted based on the reconstruction error in the corresponding label groups. Extensive experiments on MSRC and SAIAPR TC-12 datasets demonstrate the encouraging performance of our method in comparison with other baselines.

Relevance feedback is a quite effective approach to improve performance for image retrieval. Recently, active learning method has attracted much attention due to its capability of alleviating the burden of labeling in relevance feedback. However, most of the traditional studies focus on single sample selection in each feedback which needs heavy computational cost in practice. In this paper, we presents a novel batch mode active learning method for informative sample selection. Inspired by graph propagation, we consider the certainty of labels as asymmetric propagation information on graph, and formulate the correlation between labeled samples and unlabeled samples in an united scheme. Extensive experiments on publicly available data sets show that the proposed method is promising.

Selective sampling has been widely used in relevance feedback of image retrieval to alleviate the burden of labeling by selecting the most informative instances for user to label. Traditional sample selection scheme often selects a batch of instances each time and label them simultaneously, which ignores the correlation among instances and results in redundant labeling. In this paper, we propose an improved representative sampling method with certainty propagation to improve the performance of sampling. In our method, two kinds of correlations among instances are explored to reduce the redundancy in sampling. One is the correlation between labeled instances and unlabeled instances. The other is the correlation among unlabeled instances. Extensive experiments show that the proposed method achieve encouraging results.