Emotion Classification using 1D-CNN and RNN based On DEAP Dataset

Recently, emotion recognition began to be implemented in the industry and human resource field. In the time we can perceive the emotional state of the employee, the employer could gain benefits from it as they could improve the quality of decision makings regarding their employee. Hence, this subject would become an embryo for emotion recognition tasks in the human resource field. In a fact, emotion recognition has become an important topic of research, especially one based on physiological signals, such as EEG. One of the reasons is due to the availability of EEG datasets that can be widely used by researchers. Moreover, the development of many machine learning methods has been significantly contributed to this research topic over time. Here, we investigated the classification method for emotion and propose two models to address this task, which are a hybrid of two deep learning architectures: One-Dimensional Convolutional Neural Network (CNN-1D) and Recurrent Neural Network (RNN). We implement Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) in the RNN architecture, that specifically designed to address the vanishing gradient problem which usually becomes an issue in the time-series dataset. We use this model to classify four emotional regions from the valence-arousal plane: High Valence High Arousal (HVHA), High Valence Low Arousal (HVLA), Low Valence High Arousal (LVHA), and Low Valence Low Arousal (LVLA). This experiment was implemented on the well-known DEAP dataset. Experimental results show that proposed methods achieve a training accuracy of 96.3% and 97.8% in the 1DCNN-GRU model and 1DCNN-LSTM model, respectively. Therefore, both models are quite robust to perform this emotion classification task.

Managing Threats on the High Seas: The Role of Naval Bases on Reducing Maritime Piracy

Weak coastal states often lack an adequate, sustained naval presence to monitor and police their territorial waters. Unpatrolled waters, both territorial and otherwise, may provide pirates with substantial financial opportunities that go far beyond any single country. Maritime piracy costs the global economy on average USD 24 billion per year. This research explores the impact of naval bases on acts of piracy to determine if naval presence can decrease the likelihood of piracy. To examine this important economic and national security issue, our research employs a zero-inflated negative binomial regression model. We also rely upon a newly constructed time-series dataset for the years 1992–2018. Our study shows that the presence of naval bases is essential in helping maritime forces combat piracy. Policymakers searching for options to combat piracy should find the results of this study especially useful in creating prescriptive approaches that aid in solving offshore problems.

SCMFTS: Scalable and Distributed Complexity Measures and Features for Univariate and Multivariate Time Series in Big Data Environments

Time series data are becoming increasingly important due to the interconnectedness of the world. Classical problems, which are getting bigger and bigger, require more and more resources for their processing, and Big Data technologies offer many solutions. Although the principal algorithms for traditional vector-based problems are available in Big Data environments, the lack of tools for time series processing in these environments needs to be addressed. In this work, we propose a scalable and distributed time series transformation for Big Data environments based on well-known time series features (SCMFTS), which allows practitioners to apply traditional vector-based algorithms to time series problems. The proposed transformation, along with the algorithms available in Spark, improved the best results in the state-of-the-art on the Wearable Stress and Affect Detection dataset, which is the biggest publicly available multivariate time series dataset in the University of California Irvine (UCI) Machine Learning Repository. In addition, SCMFTS showed a linear relationship between its runtime and the number of processed time series, demonstrating a linear scalable behavior, which is mandatory in Big Data environments. SCMFTS has been implemented in the Scala programming language for the Apache Spark framework, and the code is publicly available.

Deep Learning Based Models: Basic LSTM, Bi LSTM, Stacked LSTM, CNN LSTM and Conv LSTM to Forecast Agricultural Commodities Prices

The literature argues that an accurate price prediction of agricultural goods is a quintessence to assure a good functioning of the economy all over the world. Research reveals that studies with application of deep learning in the tasks of agricultural price forecast on short historical agricultural prices data are very scarce and insist on the use of different methods of deep learning to predict and to this reaction of filling the gap, this study employs five versions of LSTM deep learning techniques for the task of five agricultural commodities prices prediction on univariate time series dataset of Rice, Wheat, Gram, Banana, and Groundnut spanning January 2000 to July 2020. The study obtained good forecasting results for all five commodities employing all the five LSTM models. The study validated the results with lower values of error metrics, MAE, MAPE, MSE, and RMSE and two paired t-test with hypothesis and confidence level of 95% as a measure of robustness. The study predicted the one month ahead future price for all the five commodities and compared it with actual prices using said LSTM models and obtained promising results.

On The Effect of Preprocessing Techniques For Evapotranspiration Estimation Using Soft Computing Methods

Implementing a reliable computational model for predicting the reference evapotranspiration (ET0) process is essential for several agricultural and hydrological applications, especially for the rural water resource systems, water use allocations, utilization and demand assessments, and the management of irrigation systems. In this research, two artificial intelligence (AI) models, artificial neural network (ANN) and model tree (MT), were investigated for modelling ET0. To validate model performance, five climatic stations such as Urmia, Mahabad, Takab, Khoy, and sardasht in West Azerbaijan Province of Iran. In the next step and to improve the model's accuracy, a novel preprocessing algorithm, ensemble empirical mode decomposition (EEMD), was coupled with those AI models to remove the trends or noise in the time series dataset. The extracted results indicated that the EEMD-MT model for all five stations outperformed other standalone and hybrid models.

Uni-Variate and Multi-Variate Short-Term Household Electricity Consumption Prediction Using Machine Learning Technique

Background: Electricity consumption prediction plays an important role in conservation, development, and future planning. Accurate prediction model has various field applications in real-life scenarios, future electricity demand estimation, performance evaluation of current time, fault detection, efficient energy production, resource-saving, and many more. In this paper, a CNN based short term building electricity consumption prediction model is developed and tested for two different types of datasets that can perform weekly prediction. Two different datasets are used to check how the algorithm behaves on different datasets i.e., what are the impacts dataset has on prediction accuracy. Errors were calculated using MAE and RMSE. Objective: The objective of the study is to develop an electricity consumption prediction (ECP) model for a univariate and multivariate dataset using CNN and LSTM network and to find that how the correlation and independency of features affect the electricity prediction task. Methods: The proposed electricity consumption model is built using the deep CNN andLSTM network and is trained and tested using the univariate and multivariate time series dataset thus the two experiments have been performed and are named as U-ECPCL (Univariate- Electricity Consumption Prediction using CNN and LSTM) and M-ECPCL (Multivariate- Electricity Consumption Prediction using CNN and LSTM) respectively. Results: The model predicts accurately with few errors with MAE of 0.251 and RMSE of 0.66 for univariate dataset and MAE of 4.36 and RMSE of 11.53 for a multivariate dataset. Conclusion: The model predicts accurately with few errors and if the prediction error of univariate and multivariate are compared then it is concluded that the univariate model outperforms the multivariate model.

The Problem of Fairness in Synthetic Healthcare Data

Access to healthcare data such as electronic health records (EHR) is often restricted by laws established to protect patient privacy. These restrictions hinder the reproducibility of existing results based on private healthcare data and also limit new research. Synthetically-generated healthcare data solve this problem by preserving privacy and enabling researchers and policymakers to drive decisions and methods based on realistic data. Healthcare data can include information about multiple in- and out- patient visits of patients, making it a time-series dataset which is often influenced by protected attributes like age, gender, race etc. The COVID-19 pandemic has exacerbated health inequities, with certain subgroups experiencing poorer outcomes and less access to healthcare. To combat these inequities, synthetic data must “fairly” represent diverse minority subgroups such that the conclusions drawn on synthetic data are correct and the results can be generalized to real data. In this article, we develop two fairness metrics for synthetic data, and analyze all subgroups defined by protected attributes to analyze the bias in three published synthetic research datasets. These covariate-level disparity metrics revealed that synthetic data may not be representative at the univariate and multivariate subgroup-levels and thus, fairness should be addressed when developing data generation methods. We discuss the need for measuring fairness in synthetic healthcare data to enable the development of robust machine learning models to create more equitable synthetic healthcare datasets.

Fractality of Tics as a Quantitative Assessment Tool for Diagnosis of Tourette Syndrome

Tics manifest as brief, purposeless, and involuntary movements or noises that can be suppressed temporarily with effort. In 1998, Peterson and Leckman (P&L) hypothesized that the chaotic temporal nature of tics could possess an inherent fractality, that is, have neighbor-to-neighbor correlation at all levels of time scale. However, demonstrating this phenomenon has eluded researchers for more than two decades, primarily because of the challenges associated with estimating the scale-invariant, power law exponent-called the fractal dimension Df-from a fractional Brownian noise. Here, we confirm P&L's hypothesis and establish the fractality of tics by examining year-long tic time series dataset of children diagnosed with Tourette syndrome using one-dimensional random walk models. We find that Df increases from ~1.4 to 1.75 in order of decreasing tic severity, and is correlated with the conventional YGTTS total tic score (TTS) clinical measure (p-value = 0.03). We demonstrate Df to be a sensitive parameter in examining the effect of several tic suppression conditions on the tic time series. Our findings pave the way for utilizing the fractal nature of tics as a quantitative tool for estimating tic severity and treatment effectiveness, as well as a marker for differentiating typical from functional tics.

Predicting Customer Behavior Using Prophet Algorithm In A Real Time Series Dataset

Customer Relationship Management is important in analyzing business performance. Predicting customer buying behavior enables the business to better address their customers and enhance service level and overall profit. This paper focuses on proposing a model that predicts future period sales in a real retail department store with low prediction error rate, and it also discovers the main sales trends over time. A model based on the Prophet algorithm is implemented and modified according to different parameters in order to lower the prediction error. The modifications consisted of the insertion of a new seasonality pattern, changes in the Fourier order of the existing and the new seasonality pattern, inclusion of the holiday data, and parameterizing its impact. The performance of the standard and modified model is evaluated in terms of the MAE (mean absolute error) and MAPE (mean absolute percentage error). The standard and the modified model were tested on a real dataset consisting of the sales between 2011-2019 in a department store of a shopping center in Albania. Implementation results show that the MAE in sales prediction for the modified model is reduced, while the MAPE in sales prediction for the modified model was measured for prediction periods. The implementation results indicate a comparable or evenbetter performance than the standard model. different