Explore the vast range of titles available.

Over a million people die every year as a result of road crashes. A significant proportion of all road fatalities are related to vulnerable road users, including motorcyclist and pedestrian. Artificial intelligent can proactively act to identify those drivers as higher risk of crashes. However, an accurate predictive model is essential for the safety of the road users. High accuracy could be achieved by implementing a reliable method with an optimal architecture and structure being able to identify and flag the risky drivers before a crash occurs. Over the last decade, extensive research has been conducted to achieve a higher accuracy by adjusting the depth of various machine learning algorithms or staking different processing layers. The literature highlights that having a higher algorithm complexity or blindly stacking up layers would not necessarily enhance the model accuracy. Despite the extensive efforts being made about the impacts of depth and various processing layers in the literture, not much research has been conducted on transportation problems to investigate the importance of those factors on the accuracy of crash prediction. In this context, this study aims to implement and compare various deep learning architectures in predicting motorists' crash severity. Various processing layers and depths are considered and compared. Long short‐term memory (LSTM) models have extensively been used in the literature for different types of sequential datasets. However, a comprehensive application of this method for non‐sequential data is still missing. Different processing layers of LSTM and deep neural network (DNN)‐based models with various depths and combinations are here considered and compared. The results indicate that a simple LSTM outperforms an LSTM model with higher depths, and a model with DNN stacked on top of the LSTM model. This study discusses in detail the methodological approach to stacking various layers and hyperparameters tuning. This study aims to implement and compare various deep learning architectures in predicting motorists' crash severity. Various processing layers and depths were considered and compared. This study discusses in detail the methodological approach to stacking various layers and hyperparameters tuning.

We prove that a subtle but substantial bias exists in a common measure of the conditional dependence of present outcomes on streaks of past outcomes in sequential data. The magnitude of this streak selection bias generally decreases as the sequence gets longer, but increases in streak length, and remains substantial for a range of sequence lengths often used in empirical work. We observe that the canonical study in the influential hot hand fallacy literature, along with replications, are vulnerable to the bias. Upon correcting for the bias, we find that the longstanding conclusions of the canonical study are reversed.

This paper proposes a method that arranges evaluation targets based on their complex sequential data. The data is composed of numerical one and text one. This method focuses on both change ratios of the numerical sequential data and the occupation ratios of the text sequential data. It generates a ranking model of the evaluation targets. The model can extract important evaluation targets. This paper applies the method to the data composed of stock price information and news articles. The former one corresponds to the numerical sequential data and the latter one corresponds to the text sequential data. Lastly, this paper compares the method with a method based on random selection and shows the effect of the proposed method.

Data assimilation (DA) holds considerable potential for improving hydrologic predictions as demonstrated in numerous research studies. However, advances in hydrologic DA research have not been adequately or timely implemented in operational forecast systems to improve the skill of forecasts for better informed real-world decision making. This is due in part to a lack of mechanisms to properly quantify the uncertainty in observations and forecast models in real-time forecasting situations and to conduct the merging of data and models in a way that is adequately efficient and transparent to operational forecasters. The need for effective DA of useful hydrologic data into the forecast process has become increasingly recognized in recent years. This motivated a hydrologic DA workshop in Delft, the Netherlands in November 2010, which focused on advancing DA in operational hydrologic forecasting and water resources management. As an outcome of the workshop, this paper reviews, in relevant detail, the current status of DA applications in both hydrologic research and operational practices, and discusses the existing or potential hurdles and challenges in transitioning hydrologic DA research into cost-effective operational forecasting tools, as well as the potential pathways and newly emerging opportunities for overcoming these challenges. Several related aspects are discussed, including (1) theoretical or mathematical aspects in DA algorithms, (2) the estimation of different types of uncertainty, (3) new observations and their objective use in hydrologic DA, (4) the use of DA for real-time control of water resources systems, and (5) the development of community-based, generic DA tools for hydrologic applications. It is recommended that cost-effective transition of hydrologic DA from research to operations should be helped by developing community-based, generic modeling and DA tools or frameworks, and through fostering collaborative efforts among hydrologic modellers, DA developers, and operational forecasters.

Many applications, such as intermittent data assimilation, lead to a recursive application of Bayesian inference within a Monte Carlo context. Popular data assimilation algorithms include sequential Monte Carlo methods and ensemble Kalman filters (EnKFs). These methods differ in the way Bayesian inference is implemented. Sequential Monte Carlo methods rely on importance sampling combined with a resampling step, while EnKFs utilize a linear transformation of Monte Carlo samples based on the classic Kalman filter. While EnKFs have proven to be quite robust even for small ensemble sizes, they are not consistent since their derivation relies on a linear regression ansatz. In this paper, we propose another transform method, which does not rely on any a priori assumptions on the underlying prior and posterior distributions. The new method is based on solving an optimal transportation problem for discrete random variables. [PUBLICATION ABSTRACT]

Federated Learning (FL) and Split Learning (SL) are privacy-preserving Machine-Learning (ML) techniques that enable training ML models over data distributed among clients without requiring direct access to their raw data. Existing FL and SL approaches work on horizontally or vertically partitioned data and cannot handle sequentially partitioned data where segments of multiple-segment sequential data are distributed across clients. In this paper, we propose a novel federated split learning framework, FedSL, to train models on distributed sequential data. The most common ML models to train on sequential data are Recurrent Neural Networks (RNNs). Since the proposed framework is privacy preserving, segments of multiple-segment sequential data cannot be shared between clients or between clients and server. To circumvent this limitation, we propose a novel SL approach tailored for RNNs. A RNN is split into sub-networks, and each sub-network is trained on one client containing single segments of multiple-segment training sequences. During local training, the sub-networks on different clients communicate with each other to capture latent dependencies between consecutive segments of multiple-segment sequential data on different clients, but without sharing raw data or complete model parameters. After training local sub-networks with local sequential data segments, all clients send their sub-networks to a federated server where sub-networks are aggregated to generate a global model. The experimental results on simulated and real-world datasets demonstrate that the proposed method successfully trains models on distributed sequential data, while preserving privacy, and outperforms previous FL and centralized learning approaches in terms of achieving higher accuracy in fewer communication rounds.

This study addresses challenges related to privacy issues in utilizing medical data, particularly the protection of personal information. To overcome this obstacle, the research focuses on data synthesis using real-world time-series generative adversarial networks (RTSGAN). A total of 53,005 data were synthesized using the dataset of 15,799 patients with colorectal cancer. The results of the quantitative evaluation of the synthetic data’s quality are as follows: the Hellinger distance ranged from 0 to 0.25; the train on synthetic, test on real (TSTR) and train on real, test on synthetic (TRTS) results showed an average area under the curve of 0.99 and 0.98; a propensity mean squared error was 0.223. The synthetic and real data were similar in the qualitative methods including t-SNE and histogram analyses. The application of synthetic data in predicting five-year survival in colorectal cancer patients demonstrates comparable performance to models based on real data. This study employs distance to closest records and membership inference test to assess potential privacy exposure, revealing minimal risk. This study demonstrated that it is feasible to synthesize medical data, including time-series data, using the RTSGAN, and the synthetic data can be evaluated to accurately reflect the characteristics of real data through quantitative and qualitative methods as well as by utilizing real-world artificial intelligence models.

The rapid evolution of IoT environment in medical and industrial applications has led to an increase in network vulnerabilities, making an intrusion detection system a critical requirement. Existing methods often struggle in capturing complex and irregular patterns from dynamic intrusion data, making them not suitable for different IoT applications. To address these limitations, this work proposes STID-Net that integrated customized convolutional kernels for spatial feature extraction and LSTM layers for temporal sequence modelling. Unlike traditional models, STID-Net has an improved ability to identify irregular patterns in dynamic datasets. This work is also equipped with an attention mechanism for enhancing the detection of long-term dependencies in intrusion patterns. The STID-Net is also experimented with the MBGD and SGD optimizers, and we are satisfied with the performance of the SGD optimizer in both the IoMT and IIoT datasets. The SGD optimized model provides a faster convergence and better weight adjustments for handling noisy datasets, making it robust and scalable for diverse IoT applications. This experimental work demonstrates an accuracy of 97.14% and 97.85% with the MBGD optimizer in the IoMT and IIoT datasets, while it attained 98.58% and 99.15% with SGD optimization, respectively. The proposed methodology also outperforms the standalone CNN and LSTM models incorporated with both optimizers, and the result indicates the robustness and scalability of STID-Net in medical and industrial applications.

Artificial Intelligence (AI) has recently emerged as a powerful tool with versatile applications spanning various domains. AI replicates human intelligence processes through machinery and computer systems, finding utility in expert systems, image and speech recognition, machine vision, and natural language processing (NLP). One notable area with limited exploration pertains to using deep learning models, specifically Recurrent Neural Networks (RNNs), for predicting water quality in wastewater treatment plants (WWTPs). RNNs are purpose-built for handling sequential data, featuring a feedback mechanism. However, standard RNNs may exhibit limitations in accommodating both short-term and long-term dependencies when addressing intricate time series problems. The solution to this challenge lies in adopting Long Short-Term Memory (LSTM) cells, known for their inherent memory management through a ‘forget gate’ mechanism. In general, LSTM architecture demonstrates superior performance. WWTP data represent a historical series influenced by fluctuating environmental conditions. This study employs simple RNNs and LSTM architecture to construct prediction models for effluent parameters, systematically assessing their performance through various training data scenarios and model architectures. The primary objective was to determine the most suitable WWTP dataset model. The study revealed that an epoch setting of 50 and a batch size of 100 yielded the lowest training time and root mean square error (RMSE) values for both RNN and LSTM models. Furthermore, when these models are applied to predict effluent parameters, they exhibit precise RMSE values for all parameters. The study results can be applied to detect potential upsets in WWTP operations.

The recent explosive growth in the number of smart technologies relying on data collected from sensors and processed with machine learning classifiers made the training data imbalance problem more visible than ever before. Class-imbalanced sets used to train models of various events of interest are among the main reasons for a smart technology to work incorrectly or even to completely fail. This paper presents an attempt to resolve the imbalance problem in sensor sequential (time-series) data through training data augmentation. An Unrolled Generative Adversarial Networks (Unrolled GAN)-powered framework is developed and successfully used to balance the training data of smartphone accelerometer and gyroscope sensors in different contexts of road surface monitoring. Experiments with other sensor data from an open data collection are also conducted. It is demonstrated that the proposed approach allows for improving the classification performance in the case of heavily imbalanced data (the F1 score increased from 0.69 to 0.72, p<0.01, in the presented case study). However, the effect is negligible in the case of slightly imbalanced or inadequate training sets. The latter determines the limitations of this study that would be resolved in future work aimed at incorporating mechanisms for assessing the training data quality into the proposed framework and improving its computational efficiency.