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).

ObjectiveForecasting the asset prices in any market is an inseparable and critical part of research on markets. Obtaining a credible prediction of an asset's potential future price provides valuable information for investors. Besides that, being confronted with real estate price shocks and price fluctuations in alternative markets, determining the best time for the investment may come up as a big challenge for any investor. Tracing the trend of housing price changes in Iran shows that the average housing price has had a general trend close to other prices and price indices; but the significant issue in this market is the various growth processes and dynamics, compared to changes in other economic indicators. This dynamic gets more complex when a variety of quantitative and qualitative data from diverse types and from several markets are added to the model. This diversity along with the unstructured, stochastic, and scattered datasets makes the implementation of the model so hard to do. The main goal of this paper is to build an Artificial Intelligence model with the most flexibility in the input part for the data type heterogeneity and the lowest error in the output part as the predicted price. All the implemented models showed a high accuracy on the real data extracted from the housing market. MethodsArtificial Intelligence models have an enormous capacity to get a broad range of any kind of data and process and concurrently compile them to build a valid output. This characteristic of AI is very useful in financial models to increase the accuracy of output. Our model is based on the Recurrent Neural Networks. Due to its capability to preserve past information, the LSTM algorithm was implemented as a time series forecasting model. ResultsIn this study, using diverse types of official datasets, such as the Central Bank of Iran, we distinguish the influencing variables in the housing market and then we could predict the average housing price in Tehran. Our findings indicate that the average housing price demonstrates the strongest correlation with gold prices, foreign exchange rates, the Consumer Price Index, and market liquidity levels. Utilizing these indicators, predictions with very high accuracy were obtained. ConclusionAmong the four different models of this research, the best prediction belonged to the multivariate stacked-LSTM model, which was empowered by the highly correlated macroeconomic variables. The validation of models was done by the MAPE Indicator. Furthermore, all the results confirmed that the LSTM algorithm was highly effective in utilizing over two decades of actual housing data from Tehran to forecast future prices.

Term mismatch is the most important challege in web information retrieval. The term mismatch problem is defined as differences between user queries and contents of documents while referring to the same topic. Query expansion methods deal with term mismatch by reformulating the queries to increase their term-overlap with relevant documents. In this paper, we proposed a query expansion framework based on a deep Siamese LSTM neural network. In addition, we defined the relevant relatedness for the first time and used this concept to label pairs made from user query and candidate query. Weakly-supervised labeled pairs are utilized in training of the deep Siamese network. The trained Siamese network provides labels for testset pairs in addition to contrastive loss values. The contrastive loss value reflects the cost of pulling together similar pairs. Pairs with minimum contrastive loss values are selected and merged together to form one expanded query. Results of our tests showed that the proposed framework outperforms similar word embedding based query expansion methods.

Reliability of data transmission in wireless sensor networks (WSN) is very important in the case of high lost packet rate due to link problems or buffer congestion. In this regard, mechanisms such as middle cache points and congestion control can improve the performance of the reliability of transmission protocols when the packet is lost. On the other hand, the issue of energy consumption in this type of networks has become an important parameter in their reliability. In this paper, considering the energy constraints in the sensor nodes and the direct relationship between energy consumption and the number of transmissions made by the nodes, the system tries to reduce the number of transmissions needed to send a packet from source to destination as much as possible by optimal selection of the cache points and packet caching. In order to select the best cache points, the information extracted from the network behavior analysis by deep learning algorithm has been used. In the training phase, long-short term memory (LSTM) capabilities as an example of recurrent neural network (RNN) deep learning networks to learn network conditions. The results show that the proposed method works better in examining the evaluation criteria of transmission costs, end-to-end delays, cache use and throughput.