Pub Date : 2019-11-01DOI: 10.1109/IWCIA47330.2019.8955094
Takeshi Kawachi, J. Kushida, Akira Hara, T. Takahama
Real world problems are often formularized as constrained optimization problems (COPs). Constraint handling techniques are important for efficient search, and various approaches such as penalty methods or feasibility rules have been studied. The penalty methods deal with a single fitness function by combining the objective function value and the constraint violation with a penalty factor. Moreover, the penalty factor can be flexibly adapted by feeding back information on search process in adaptive penalty methods. However, keeping the good balance between the objective function value and the constraint violation is very difficult. In this paper, we propose a new adaptive penalty method with balancing the objective function value and the constraint violation and examine its effectiveness. L-SHADE is adopted as a base algorithm to evaluate search performance, and the optimization results of 28 benchmark functions provided by the CEC 2017 competition on constrained single-objective numerical optimizations are compared with other methods. In addition, we also examine the behavioral difference between proposed method and the conventional adaptive penalty method.
{"title":"Efficient Constraint Handling based on the Adaptive Penalty Method with Balancing the Objective Function Value and the Constraint Violation","authors":"Takeshi Kawachi, J. Kushida, Akira Hara, T. Takahama","doi":"10.1109/IWCIA47330.2019.8955094","DOIUrl":"https://doi.org/10.1109/IWCIA47330.2019.8955094","url":null,"abstract":"Real world problems are often formularized as constrained optimization problems (COPs). Constraint handling techniques are important for efficient search, and various approaches such as penalty methods or feasibility rules have been studied. The penalty methods deal with a single fitness function by combining the objective function value and the constraint violation with a penalty factor. Moreover, the penalty factor can be flexibly adapted by feeding back information on search process in adaptive penalty methods. However, keeping the good balance between the objective function value and the constraint violation is very difficult. In this paper, we propose a new adaptive penalty method with balancing the objective function value and the constraint violation and examine its effectiveness. L-SHADE is adopted as a base algorithm to evaluate search performance, and the optimization results of 28 benchmark functions provided by the CEC 2017 competition on constrained single-objective numerical optimizations are compared with other methods. In addition, we also examine the behavioral difference between proposed method and the conventional adaptive penalty method.","PeriodicalId":139434,"journal":{"name":"2019 IEEE 11th International Workshop on Computational Intelligence and Applications (IWCIA)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129923927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-09-30DOI: 10.1109/IWCIA47330.2019.8955072
T. Ichimura, Shin Kamada
AffectNet contains more than 1,000,000 facial images which manually annotated for the presence of eight discrete facial expressions and the intensity of valence and arousal. Adaptive structural learning method of DBN (Adaptive DBN) is positioned as a top Deep learning model of classification capability for some large image benchmark databases. The Convolutional Neural Network and Adaptive DBN were trained for AffectNet and classification capability was compared. Adaptive DBN showed higher classification ratio. However, the model was not able to classify some test cases correctly because human emotions contain many ambiguous features or patterns leading wrong answer which includes the possibility of being a factor of adversarial examples, due to two or more annotators answer different subjective judgment for an image. In order to distinguish such cases, this paper investigated a re-learning model of Adaptive DBN with two or more child models, where the original trained model can be seen as a parent model and then new child models are generated for some misclassified cases. In addition, an appropriate child model was generated according to difference between two models by using KL divergence. The generated child models showed better performance to classify two emotion categories: ‘Disgust’ and ‘Anger’.
AffectNet包含超过1,000,000张面部图像，这些图像手动注释了8个离散的面部表情以及效价和唤醒的强度。DBN的自适应结构学习方法(Adaptive structural learning method of DBN，简称Adaptive DBN)定位为一些大型图像基准数据库分类能力的顶级深度学习模型。对卷积神经网络和自适应DBN进行了AffectNet的训练，并对其分类能力进行了比较。自适应DBN具有较高的分类率。然而，该模型无法正确分类一些测试用例，因为人类情感包含许多模糊的特征或模式，导致错误的答案，其中包括由于两个或更多注释者对图像回答不同的主观判断而成为对抗性示例因素的可能性。为了区分这种情况，本文研究了一种具有两个或多个子模型的自适应DBN再学习模型，其中原始训练的模型可以视为父模型，然后对一些错误分类的情况生成新的子模型。此外，利用KL散度根据两个模型之间的差异生成合适的子模型。生成的儿童模型在区分“厌恶”和“愤怒”两种情绪类别方面表现得更好。
{"title":"Re-learning of Child Model for Misclassified data by using KL Divergence in AffectNet: A Database for Facial Expression","authors":"T. Ichimura, Shin Kamada","doi":"10.1109/IWCIA47330.2019.8955072","DOIUrl":"https://doi.org/10.1109/IWCIA47330.2019.8955072","url":null,"abstract":"AffectNet contains more than 1,000,000 facial images which manually annotated for the presence of eight discrete facial expressions and the intensity of valence and arousal. Adaptive structural learning method of DBN (Adaptive DBN) is positioned as a top Deep learning model of classification capability for some large image benchmark databases. The Convolutional Neural Network and Adaptive DBN were trained for AffectNet and classification capability was compared. Adaptive DBN showed higher classification ratio. However, the model was not able to classify some test cases correctly because human emotions contain many ambiguous features or patterns leading wrong answer which includes the possibility of being a factor of adversarial examples, due to two or more annotators answer different subjective judgment for an image. In order to distinguish such cases, this paper investigated a re-learning model of Adaptive DBN with two or more child models, where the original trained model can be seen as a parent model and then new child models are generated for some misclassified cases. In addition, an appropriate child model was generated according to difference between two models by using KL divergence. The generated child models showed better performance to classify two emotion categories: ‘Disgust’ and ‘Anger’.","PeriodicalId":139434,"journal":{"name":"2019 IEEE 11th International Workshop on Computational Intelligence and Applications (IWCIA)","volume":"110 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115696997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-09-30DOI: 10.1109/IWCIA47330.2019.8955036
Shin Kamada, T. Ichimura
Deep learning builds deep architectures such as multi-layered artificial neural networks to effectively represent multiple features of input patterns. The adaptive structural learning method of Deep Belief Network (DBN) can realize a high classification capability while searching the optimal network structure during the training. The method can find the optimal number of hidden neurons of a Restricted Boltzmann Machine (RBM) by neuron generation-annihilation algorithm to train the given input data, and then it can make a new layer in DBN by the layer generation algorithm to actualize a deep data representation. Moreover, the learning algorithm of Adaptive RBM and Adaptive DBN was extended to the time-series analysis by using the idea of LSTM (Long Short Term Memory). In this paper, our proposed prediction method was applied to Moving MNIST, which is a benchmark data set for video recognition. We challenge to reveal the power of our proposed method in the video recognition research field, since video includes rich source of visual information. Compared with the LSTM model, our method showed higher prediction performance (more than 90% predication accuracy for test data).
{"title":"A Video Recognition Method by using Adaptive Structural Learning of Long Short Term Memory based Deep Belief Network","authors":"Shin Kamada, T. Ichimura","doi":"10.1109/IWCIA47330.2019.8955036","DOIUrl":"https://doi.org/10.1109/IWCIA47330.2019.8955036","url":null,"abstract":"Deep learning builds deep architectures such as multi-layered artificial neural networks to effectively represent multiple features of input patterns. The adaptive structural learning method of Deep Belief Network (DBN) can realize a high classification capability while searching the optimal network structure during the training. The method can find the optimal number of hidden neurons of a Restricted Boltzmann Machine (RBM) by neuron generation-annihilation algorithm to train the given input data, and then it can make a new layer in DBN by the layer generation algorithm to actualize a deep data representation. Moreover, the learning algorithm of Adaptive RBM and Adaptive DBN was extended to the time-series analysis by using the idea of LSTM (Long Short Term Memory). In this paper, our proposed prediction method was applied to Moving MNIST, which is a benchmark data set for video recognition. We challenge to reveal the power of our proposed method in the video recognition research field, since video includes rich source of visual information. Compared with the LSTM model, our method showed higher prediction performance (more than 90% predication accuracy for test data).","PeriodicalId":139434,"journal":{"name":"2019 IEEE 11th International Workshop on Computational Intelligence and Applications (IWCIA)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114531946","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：