The stock market is influenced by various factors such as national policies, economic conditions, and global events. Accurately predicting stock price changes has long been a critical challenge for investors and economists, as it can significantly reduce investment risks. Accurate forecasts can enhance investment strategies, allowing for maximized returns. However, the volatile and non-linear nature of financial markets makes this task particularly complex. Recently state space models, like Mamba, have shown promising results in sequence modeling. In this work, we apply Mamba to predict the percentage changes in daily stock closing prices. By forecasting these changes, we frame the problem as a classification task, where the goal is to determine whether the stock price will increase or not. By training models with different hyperparameters and combining them through ensemble learning, the prediction accuracy is further improved. To evaluate the model, we analyze stock movements over a series of trading days for six Nasdaq-listed companies. Our model demonstrates notable performance, achieving an average F1 score of 60.5% in predicting the direction of next-day price movement. The model generated an average profit of $4150 over 120 test transaction days with an initial capital of $10,000. These insights can help investors make more informed decisions, optimizing returns while minimizing risks.

The recent rapid developments in generative-AI research has made it exceedingly hard to distinguish artificially generated audio-visual content from real ones. As a result, reliably detecting synthetic content has become an important problem to solve. In this study, multiple CNN, FC and SVM models are trained to detect synthetic audio signals obtained by using generative-AI models. The test results show that the best accuracy scores for the CNN, FC and SVM models are 99.06, 99.15 and 98.68%, respectively. These results point out that the synthetic audio signals can be discriminated from the real ones by the trained models. Therefore, the proposed solution can be used in real-life practical applications to tackle this problem. Our analyses show that CNN models are the most suitable compared to other techniques, as FC and SVM models can also detect synthetic audios but have different inherent disadvantages.

In this article, the decision mechanism of a control system has been created by using big data and by applying an ontology model. This type of control is important in order to minimize, or even eliminate human influence in systematic industrial processes. The development presented in the article in order to make the ontology model compatible with a relational database in the decision mechanism, is a step taken to eliminate the human effect. The method used in the article aims to carry out a decision mechanism under the guidance of big data by using relational databases integrated with the ontology model. In line with this goal, the ontology model associated with relational databases with high prevalence will be able to access the continuous data required for the decision process and enrich the decision mechanism. In this study, when the necessary parameters for the decision process are obtained, dynamic threshold determination is provided by a machine learning model with these parameters. This dynamic threshold varies over various time periods, with the combination of inputs provided to the machine learning model and differences in value. Our test results state that the Decision Tree model predictions’ accuracy is 100%.

In the contemporary rotating machinery, bearings are critical and indispensable parts. Early detection of rolling bearing defects carries crucial importance, because undetected defects on the rolling bearings may end in loss of time, resources, money and even lives. In parallel to the accelerated utilization of deep learning applications in the manufacturing industry, different studies have been conducted to determine and evaluate defects on the surfaces of rolling bearings. In this study, a new system, that contains a hardware platform and software components in order to detect surface defects of the metal rolling bearings has been developed. To detect defects, optic image data of the bearings were used, and then computer vision and artificial intelligence techniques were applied to them. In the system, TC-VISION, the source of big data is the platform designed and developed using the optical camera. The results of the applied CNN algorithms performed better than the targeted values with respect to several metrics. The F1 score obtained is close to 100%. The developed system is aimed to be enhanced further in order to develop a fully automated inspection and quality control system for metal rolling bearing systems appropriate for serial production in real industrial environments.

Milling is a highly crucial machining process in the modern industry. With the recent trends of Industry 4.0, it is becoming more common to implement Artificial Intelligence (AI) methods to increase the performance of milling processes. As a significant limitation for the efficiency of the machining processes, chatter detection, and avoidance are critical. In this paper, a chatter detection method based on vibration data features for the slot milling process is proposed. This method benefits from a deep learning method, Deep Multi-Layer Perceptron (DMLP). Vibration data was acquired by attaching an accelerometer to the spindle housing during slot milling operations. Fast Fouries Transform (FFT) was applied to time-domain vibratory data. Frequency domain data achieved by FFT was investigated for labeling the occurrence of chatter. These labels were used to train the DMLP algorithm. Time-domain signal features such as root mean square, clearance factor, skewness, crest factor, and shape factor were selected as inputs for the chatter detection algorithm. Finally, validation cuttings were performed for verifying the results of the DMLP algorithm. The results prove that time-domain features can provide enough information about the chatter occurrence in slot milling operations, and the DMLP algorithm proposed in this research can successfully detect the chatter occurrence.

Internet of Things world is getting bigger everyday with new developments in all fronts. The new IoT world requires better handling of big data and better usage with more intelligence integrated in all phases. Here we present MIS-IoT (Modular Intelligent Server Based Internet of Things Framework with Big Data and Machine Learning) framework, which is "modular" and therefore open for new extensions, "intelligent" by providing machine learning and deep learning methods on "big data" coming from IoT objects, "server-based" in a service-oriented way by offering services via standart Web protocols. We present an overview of the design and implementation details of MIS-IoT along with a case study evaluation of the system, showing the intelligence capabilities in anomaly detection over real-time weather data.

In recent years, big data and Internet of Things (IoT) implementations started getting more attention. Researchers focused on developing big data analytics solutions using machine learning models. Machine learning is a rising trend in this field due to its ability to extract hidden features and patterns even in highly complex datasets. In this study, we used our Big Data IoT Framework in a weather data analysis use case. We implemented weather clustering and sensor anomaly detection using a publicly available dataset. We provided the implementation details of each framework layer (acquisition, ETL, data processing, learning and decision) for this particular use case. Our chosen learning model within the library is Scikit-Learn based k-means clustering. The data analysis results indicate that it is possible to extract meaningful information from a relatively complex dataset using our framework.

Predictive Analytics is a crucial part of a Big Data application. Lately, developers have turned their attention to deep learning models due to their huge success in various implementations. Meanwhile, there is lack of deep learning implementations in manufacturing applications due to insufficient data. This phenomenon has been slowly shifting due to the application of IoT and Industry 4.0 concept within the manufacturing industry. Streaming and batch data producing sources are becoming more and more common in the machining industry. In this paper, we propose a deep learning predictive analytics model based on the data generated by a particular machining process. The results indicate that using such a model can make very accurate predictions and can be used as part of a real-time decision-making process in the manufacturing industry. In this study, the prediction models of three crucial metrics of machining such as quality, performance and energy consumption have been developed by utilizing artificial neural networks and deep learning methods. Specific measures of quality, performance and energy consumption refer to material removal rate (MRR), surface roughness (Ra) and specific energy consumption (SEC) respectively. The control parameters of machining are selected as stepover (ae), depth of cut (ap), feed per tooth (fz) and cutting speed (Vc). In addition, variance analysis (ANOVA) has been used to examine the effects of the input parameters on the output parameters.

Social Network Analysis (SNA) has become a very important and increasingly popular topic among researchers in recent years especially after emerging Semantic Web and Big Data technologies. Social networking services such as Facebook, Google+, Twitter, etc. provide large amounts of data that can be used for social network analysis by researchers. Semantic Web technology plays an important role for collecting, merging, and aggregating social network data from heterogeneous sources more easily, robustly and in an interoperable manner. Today, data scientists use several different frameworks for querying, integrating and analyzing datasets located at different sources. Meanwhile, most of the big social data is in unstructured or semi-structured format. Big data architectures allow researchers to analyze unstructured data in a time and cost-efficient way. New approaches for SNA are needed to combine Semantic Web and Big Data technologies in order to utilize and add capabilities to existing solutions. To be able to analyze large scale social networks, algorithms should have scalable designs in order to benefit from the emerging Big Data technologies. This survey focuses on recently developed systems for SNA and summarizes the state-of-the-art technologies used by them and points out to future research directions.

Due to increasing urban population and growing number of motor vehicles, traffic congestion is becoming a major problem of the 21st century. One of the main reasons behind traffic congestion is accidents which can not only result in casualties and losses for the participants, but also in wasted and lost time for the others that are stuck behind the wheels. Early detection of an accident can save lives, provides quicker road openings, hence decreases wasted time and resources, and increases efficiency. In this study, we propose a preliminary real-time autonomous accident-detection system based on computational intelligence techniques. Istanbul City traffic-flow data for the year 2015 from various sensor locations are populated using big data processing methodologies. The extracted features are then fed into a nearest neighbor model, a regression tree, and a feed-forward neural network model. For the output, the possibility of an occurrence of an accident is predicted. The results indicate that even though the number of false alarms dominates the real accident cases, the system can still provide useful information that can be used for status verification and early reaction to possible accidents.

Many experts claim that data will be the most valuable commodity in the 21st century. At the same time, two of the most influential components of this era, Big Data and IoT are moving very fast, on a collision course with the methodologies that are associated with conventional data processing and database systems. As a result, new approaches like NoSQL databases, distributed architectures, etc. started appearing on the stage. Meanwhile, another technology, ontology and semantic data processing can be a very convenient catalyzer that might assist in smoothly providing this transformation process. In this paper, we propose a combined framework that brings Big Data, IoT, and semantic web together to build an augmented framework for this new era. We not only list the components of such a system and define the necessary bindings that needs to be integrated together, but also provide a realistic use case that demonstrates how the model can implement the desired functionality and achieve the goals of such a model.

Achieving high quality clustering is one of the most well-known problems in data mining. k-means is by far the most commonly used clustering algorithm. It converges fairly quickly, but achieving a good solution is not guaranteed. The clustering quality is highly dependent on the selection of the initial centroid selections. Moreover, when the number of clusters increases, it starts to suffer from "empty clustering". The motivation in this study is two-fold. We not only aim at improving the k-means clustering quality, but at the same time not being effected by the empty cluster issue. For achieving this purpose, we developed a hybrid model, H(EC)2S, Hybrid Evolutionary Clustering with Empty Clustering Solution. Firstly, it selects representative points to eliminate Empty Clustering problem. Then, the hybrid algorithm uses only these points during centroid selection. The proposed model combines Fireworks and Cuckoo-search based evolutionary algorithm with some centroid-calculation heuristics. The model is implemented using a Hadoop Mapreduce algorithm for achieving scalability when faced with a Big Data clustering problem. The advantages of the developed model is particularly attractive when the amount, dimensionality and number of cluster parameters tend to increase. The results indicate that considerable clustering quality performance improvement is achieved using the proposed model.