Computational Intelligence最新文献

Combining the stock prediction with portfolio optimization can improve the performance of the portfolio construction. In this article, we propose a novel portfolio construction approach by utilizing a two-stage ensemble model to forecast stock prices and combining the forecasting results with the portfolio optimization. To be specific, there are two phases in the approach: stock prediction and portfolio optimization. The stock prediction has two stages. In the first stage, three neural networks, that is, multilayer perceptron (MLP), gated recurrent unit (GRU), and long short-term memory (LSTM) are used to integrate the forecasting results of four individual models, that is, LSTM, GRU, deep multilayer perceptron (DMLP), and random forest (RF). In the second stage, the time-varying weight ordinary least square model (OLS) is utilized to combine the first-stage forecasting results to obtain the ultimate forecasting results, and then the stocks having a better potential return on investment are chosen. In the portfolio optimization, a diversified mean-variance with forecasting model named DMVF is proposed, in which an average predictive error term is considered to obtain excess returns, and a 2-norm cost function is introduced to diversify the portfolio. Using the historical data from the Shanghai stock exchange as the study sample, the results of the experiments indicate the DMVF model with two-stage ensemble prediction outperforms benchmarks in terms of return and return-risk characteristics.

Mobile edge computing (MEC) is an important research topic in the field of wireless communication and mobile computing, as it can effectively decrease the latency and energy consumption due to the trade-off between the communication and computing, where some intensive computing tasks can be offloaded to computational access points (CAPs), especially when the wireless transmission channel is in good condition. This article studies how to intelligently allocate the computing capability and wireless bandwidth among users for a cache-aided multi-terminal multi-CAP MEC network with non-ideal channel estimation, where there are $��N�$ mobile terminals and $��M�$ CAPs in the network. Each terminal has some tasks that need to be computed in a fast and efficient way. For such a system, we first design the system by jointly considering the computing capability and wireless bandwidth allocation, where the computing and communication delay is used as the performance of metric. To optimize the system performance, we then employ deep deterministic policy gradient to learn an effective strategy on the allocation of computing capability and wireless bandwidth, in order to decrease the system delay as much as possible. Simulations are finally conducted to show the superiority of the proposed studies in this article, especially about the advantages from cache.

The rapid development of new technologies such as artificial intelligence and big data analysis requires the simultaneous development of cloud computing technology. The application of IoT-to-cloud setting has been fully applied in various industry sectors, such as sensor-cloud system which is composed of wireless sensor network and cloud computing technology. With the increasing amount and types of collected data, companies need to reduce the dimension of massive data in cloud servers for obtaining data analysis reports rapidly. Due to frequent cloud server data leaks, companies must adequately protect the privacy of some confidential data. To this end, we designed a dimension reduction method for ciphertext data in the sensor-cloud system based on the CKKS encryption scheme, principal component analysis (PCA) and linear discriminant analysis (LDA) dimension reduction algorithm. As data cannot be directly calculated using traditional PCA and LDA algorithm after encryption, we add some interactive operations and iterative calculations to replace some steps in traditional algorithms. Finally, we select the classification dataset IRIS which is commonly used in machine learning, and screen out the best encryption and calculation parameters, and efficiently realize the dimension reduction method of ciphertext data through a large number of experiments.

Traffic detection (including lane detection and traffic sign detection) is one of the key technologies to realize driving assistance system and auto drive system. However, most of the existing detection methods are designed based on single-modal visible light data, when there are dramatic changes in lighting in the scene (such as insufficient lighting in night), it is difficult for these methods to obtain good detection results. In view of multi-modal data can provide complementary discriminative information, based on the YoLoV5 model, this paper proposes a multi-modal fusion YoLoV5 network, which consists of three key components: the dual stream feature extraction module, the correlation feature extraction module, and the self-attention fusion module. Specifically, the dual stream feature extraction module is used to extract the features of each of the two modalities. Secondly, input the features learned from the dual stream feature extraction module into the correlation feature extraction module to learn the features with maximum correlation. Then, the extracted maximum correlation features are used to achieve information exchange between modalities through a self-attention mechanism, and thus obtain fused features. Finally, the fused features are inputted into the detection layer to obtain the final detection result. Experimental results on different traffic detection tasks can demonstrate the superiority of the proposed method.

Script identification of text in scene images has attracted massive attention recently. However, the existing techniques primarily emphasize on scripts where data are available abundantly, such as English, European, or East Asian. Although these methods are robust in dealing with high-resource data, how these techniques will work on low-resource scripts has yet to be discovered. For example, in India, there is a disparity among the text scripts across the country's demographic. To bridge the research gap for resource-constraint script identification, we present a few-shot learning network called the TextScriptFSLNet. This network does not require huge training data while achieving state-of-the-art performance on benchmark datasets. Our proposed method acts in accordance with a $��C�$-way $��K�$-shot paradigm by splitting the train set as support and query set, respectively. The support set learns representative knowledge of each class and creates its prototypes. We use multi-kernel spatial attention fused 2-layer convolutional neural network and averaging technique to generate the prototype of each class. This spatial attention aids in grasping important information in an image and enriches the feature representation. To the best of our knowledge, the proposed work is the first of its kind in the scene text understanding domain. Additionally, we created a dataset called Indic-FSL2023 comprising 10 of the 22 officially recognized Indian scripts. The proposed method achieves the highest accuracy among the tested methods on the newly created Indic-FSL2023. Experiments are also conducted on MLe2e to demonstrate its versatility. Furthermore, we also showed how our proposed model performed concerning illumination changes and blur on scene text script images.

In many fields, like aspect category detection (ACD) in aspect-based sentiment analysis, it is necessary to label each instance with more than one label at the same time. This study tackles the multilabel classification problem in the ACD task for the Arabic language. For this purpose, we used Arabic hotel reviews from the SemEval-2016 dataset, comprising 13,113 annotated tuples provided for training (10,509) and testing (2,604). To extract valuable information, we first propose specific data preprocessing. Then, we suggest using the dynamic weighted loss function and a data augmentation method to fix the problem with this dataset's imbalance. Using two possible approaches, we develop new ways to find different categories of things in a review sentence. The first is based on classifier chains using machine learning models. The second is based on transfer learning using pretrained AraBERT fine-tuning for contextual representation. Our findings show that both approaches outperformed the related works for ACD on the Arabic SemEval-2016. Moreover, we observed that AraBERT fine-tuning performed much better and achieved a promising $��{�F�}_{�1�}�$-score of $��68�.�02�\%�$.

Stock prices forecasting has always been a challenging task. Although many research projects adopt machine learning and deep learning algorithms to address the problem, few of them pay attention to the varying degrees of dependencies between stock prices. In this paper we introduce a hybrid model that improves stock price prediction by emphasizing the dependencies between adjacent stock prices. The proposed model, ResNLS, is mainly composed of two neural architectures, ResNet and LSTM. ResNet serves as a feature extractor to identify dependencies between stock prices across time windows, while LSTM analyses the initial time-series data with the combination of dependencies which considered as residuals. In predicting the SSE Composite Index, our experiment reveals that when the closing price data for the previous five consecutive trading days is used as the input, the performance of the model (ResNLS-5) is optimal compared to those with other inputs. Furthermore, ResNLS-5 outperforms vanilla CNN, RNN, LSTM, and BiLSTM models in terms of prediction accuracy. It also demonstrates at least a 20% improvement over the current state-of-the-art baselines. To verify whether ResNLS-5 can help clients effectively avoid risks and earn profits in the stock market, we construct a quantitative trading framework for back testing. The experimental results show that the trading strategy based on predictions from ResNLS-5 can successfully mitigate losses during declining stock prices and generate profits in the periods of rising stock prices.

The utilization of distributed multi-agent unmanned aerial vehicles (UAVs) for computing tasks in remote areas has gained significant traction in recent years due to their adaptability and capability to access hard-to-reach regions that are inaccessible to ground-based methods. However, establishing wireless communication between UAVs and ground-based data sources in remote areas presents considerable challenges, particularly when UAVs are in motion. To tackle this challenge, this article investigates a cache-aided relaying system in the presence of UAVs, wherein a ground-based decode-and-forward relay equipped with cache space is deployed to facilitate wireless communication between UAVs and a central data source. Within the scope of this system, we first analyze the probability of transmission outage, providing an analytical expression for performance evaluation. We commence with the case of a single stationary UAV, subsequently expanding to multiple stationary UAVs, and ultimately incorporating multiple dynamic UAVs. Subsequently, we enhance the system performance by minimizing the outage probability through efficient power resource allocation among users. By means of mathematical modeling and simulations, this research examines the influence of various factors, including the cache size at the relay and the working mode of the UAV, on the system performance. Finally, simulations are conducted to validate the proposed analysis.

Emotional recognition in conversations (ERC) is increasingly being applied in various IoT devices. Deep learning-based multimodal ERC has achieved great success by leveraging diverse and complementary modalities. Although most existing methods try to adopt attention mechanisms to fuse different information, these methods ignore the complementarity between modalities. To this end, the joint cross-attention model is introduced to alleviate this issue. However, multi-scale feature information on different modalities is not utilized. Moreover, the context relationship plays an important role in feature extraction in the expression recognition task. In this paper, we propose a novel joint hierarchical graph convolution network (JHGCN) which exploits different layer features and context relationships for facial expression recognition based on audio-visual (A-V) information. Specifically, we adopt different deep networks to extract features from different modalities individually. For V modality, we construct V graph data based on patch embeddings which are extracted from the transformer encoder. Moreover, we embed the graph convolution which can leverage the intra-modality relationships with the transformer encoder. Then, the deep feature from different layers is fed to the hierarchical fusion module to enhance feature representation. At last, we use the joint cross-attention mechanism to exploit the complementary inter-modality relationships. To validate the proposed model, we have conducted various experiments on the AffWild2 and CMU-MOSI datasets. All results confirm that our proposed model achieves highly promising performance compared to the joint cross-attention model and other methods.