autoencoder

Text clustering is a method for separating specific information from textual data and can even classify text according to topic and sentiment, which has drawn much interest in recent years. Deep clustering methods are especially important among clustering techniques because of their high accuracy. These methods include two main components: dimensionality reduction and clustering. Many earlier efforts have employed autoencoder for dimension reduction; however, they are unable to lower dimensions based on manifold structures, and samples that are like one another are not necessarily placed next to one another in the low dimensional. In the paper, we develop a Deep Text Clustering method based on a local Manifold in the Autoencoder layer (DCTMA) that employs multiple similarity matrices to obtain manifold information, such that this final similarity matrix is obtained from the average of these matrices. The obtained matrix is added to the bottleneck representation layer in the autoencoder. The DCTMA's main goal is to generate similar representations for samples belonging to the same cluster; after dimensionality reduction is achieved with high accuracy, clusters are detected using an end-to-end deep clustering. Experimental results demonstrate that the suggested method performs surprisingly well in comparison to current state-of-the-art methods in text datasets.

In recent years, the use of Internet of Things in societies has grown widely. On the other hand, a new technology called Software Defined Networks has been proposed to solve the challenges of the Internet of Things. The security problems in these Software Defined Networks and the Internet of Things have made SDN-IoT security one of the most important concerns. On the other hand, the use of intelligent algorithms has been an opportunity that these algorithms have been able to make significant progress in various cases such as image processing and disease diagnosis. Of course, intrusion detection systems for SDN-IoT environment still face the problem of high false alarm rate and low accuracy.In this article, a new hybrid method based on intelligent algorithms is proposed. The proposed method integrates the monitoring algorithms of frequent return gate and unsupervised k-means classifier in order to obtain suitable results in the field of intrusion detection. The simulation results show that the proposed method, by using the advantages of each of the integrated algorithms and covering each other's disadvantages, has more accuracy and a lower false alarm rate than other methods such as the Hamza method. Also, the proposed method has been able to reduce the false alarm rate to 1.1% and maintain the accuracy at around 99%.

Diagnosing the type of cancer, which is called the subtype, is very important in determining the treatment process. This paper focuses on the diagnose of the four subtypes of the brain cancer. Disease subtype diagnosis can be modeled as a classification problem. Due to the significant progress made in bioinformatics in extracting genetic information from the human body, recently this information is widely used in the representing of patients in machine learning. In this paper, three types of genetic information including mRNA, miRNA and DNA methylation are used.It should be noted that combining different information sources in the form of multimodal data instead of using a single information source increases the accuracy of information classification. To extract more desirable features from the original genetic data, auto-encoder has been used so that the features extracted from auto-encoder are concatenated to the original genetic data.Random forest has performed well as a classifier in classifying patients based on genetic information. By extending deep methods in neural networks and their good performance, a version of deep random forest with layered structure has been proposed. The deep random forest has the advantage that has a limited number of parameters and lower computational complexity in addition to the optimal performance in information classification. In this paper, deep random forest is used to determine the subtype of a special type of brain cancer. The experiment results show the desired performance of the proposed method.

Today, social networks play a crucial role in disseminating information worldwide. Twitter is one of the most popular social networks, with 500 million tweets sent on a daily basis. The popularity of this network among users has led spammers to exploit it for distributing spam posts. This paper employs a combination of machine learning methods to identify spam at the tweet level. The proposed method utilizes a feature extraction framework in two stages. In the first stage, Stacked Autoencoder is used for feature extraction, and in the second stage, the extracted features from the last layer of Stacked Autoencoder are fed into the softmax layer for prediction. The proposed method is compared and evaluated against some popular methods on the Twitter Spam Detection corpus using accuracy, precision, recall, and F1-score metrics. The research results indicate that the proposed method achieves a detection of 78.1%. Overall, the proposed method, using the majority voting approach with a hard selection in ensemble learning, outperforms CNN, LSTM, and SCCL methods in identifying spam tweets with higher accuracy.

Since the formation and Direct solving the governing equations requires high time and computational cost, this study seeks to provide an equation free model based on deep learning algorithm that simulates steady state heat transfer in two-dimensional space and a relatively large size using order reduction method. Principal component analysis is a linear method and autoencoder is a nonlinear methods. The results of comparing their performance on different data sets showed that in reducion of order to very low dimensions, autoencoder and in reducion of order to very high dimensions, principal component analysis has a higher accuracy. Of course, the number of dimensions to order reduction and the characteristics of the data set such as size and number of dimensions of the data will affect the accuracy of the dimensional reduction. These two methods were used to order reduction of thermal data in order to faster simulate the phenomenon of Steady State Heat Transfer and were compared with a model based on convolutional neural network with a number of layers and multiple filters. The results showed that the models based on order reduction methods have much less computational volume and simulation time, and the outputs obtained from them, especially the model based on the autoencoder method, have a much higher accuracy.

Recently, a number of Extreme Learning Machine (ELM) based training algorithms have been introduced for training deep neural network structures. ELM based Auto-Encoder (ELM-AE) is one such algorithm that has been used for making multilayer structures and tuning parameters of each layer. In a simple ELM-AE training algorithm, the weights of the first layer are initialized randomly. This issue is a leading factor in producing reconstruction error. The frequent use of ELM-AE in deep network layers results in propagating such errors through deep structures and in decreasing performance as a consequent. In this paper, we introduce a multilayer structure and a new learning algorithm to train it that prevents error propagation. In order to boost the performance of the model, the parameters in the first layer are initialized by a novel type of ELM-AE called Repeated-AE (RAE) rather than by a random selection method. This RAE-based technique determines the parameters in the first layer far better than do the other ELM-AE existed methods. Next, a single hidden layer ELM is applied for handling the classification task. Experimental results for data classification show that the proposed method outperforms some other methods in terms of the average accuracy over all datasets by amounts of 4%, 26%, 17% and 31%. Eventually, so as to verify the performance of the proposed multilayer ELM-AE in application, we used this model to reconstruct images. The reconstructed images obtained by our approach appeared visually a lot better compared to those obtained by the other methods do.

Anomaly detection means detecting samples that are different from the normal samples in the dataset. One of the great challenges in this area is finding labeled data, especially for the abnormal categories. In this paper, we propose a method that uses normal data to detect anomalies. This method is based on established neural networks which are called automated encoder and are considered in deep learning studies. An automated encoder reproduces its input as output and reconstruction deviation to rate anomalies. We have used LSTM blocks to construct encoder instead of using ordinary neurons. In fact, these blocks are a category of recurring neural networks that are specialized in discovering and fetching time and proximity dependencies. The result of employing an automated encoder using LSTM blocks to detect point anomalies shows that this approach has been promising and successful in extracting the normal data’s internal model and also detecting anomalous data. The AUC factor of the model, in almost all cases, is better than the AUC of an ordinary automated encoder and One Class Support Vector Machine (OC-SVM).

In future smart grids, it's imperative to know the price of electricity market to guide the behavior of consumers and suppliers. This paper presents a hybrid approach for mid-term electricity price forecasting based on support vector machine and neural networks. In this method, at first, the price upper bound is considered. Then, the training set is divided into two parts including normal price and price spikes. Feature extraction applies on input data sets using stacked auto-encoders and a prediction model trained using each training set. Support Vector Machine (SVM) models with different kernel functions and a two layered feed-forward neural network were trained and tested with the proposed method. Simulation results using the proposed method show that this method has a significant effect on the speed of model training and improves forecasting accuracy.