Multiobjective multitasking optimization assisted by multidirectional prediction method

Multiobjective multitasking optimization (MTO) is an emerging research topic in the field of evolutionary computation, which has attracted extensive attention, and many evolutionary multitasking (EMT) algorithms have been proposed. One of the core issues, designing an efficient transfer strategy, has been scarcely explored. Keeping this in mind, this paper is the first attempt to design an efficient transfer strategy based on multidirectional prediction method. Specifically, the population is divided into multiple classes by the binary clustering method, and the representative point of each class is calculated. Then, an effective prediction direction method is developed to generate multiple prediction directions by representative points. Afterward, a mutation strength adaptation method is proposed according to the improvement degree of each class. Finally, the predictive transferred solutions are generated as transfer knowledge by the prediction directions and mutation strengths. By the above process, a multiobjective EMT algorithm based on multidirectional prediction method is presented. Experiments on two MTO test suits indicate that the proposed algorithm is effective and competitive to other state-of-the-art EMT algorithms.

Using Zero-Offset VSP to Identify Interbed Multiples Generators, A Case Study from South Oman

The presence of interbed multiples is a serious concern in surface seismic processing and interpretation. Its impact is huge especially if they are masking the desirable primary reflections such as the targeted reservoirs area. The conventional demultiple methodologies such as stacking, and deconvolution often fail to suppress all the interbed multiples. Therefore, a need for other measurement is crucial to eliminate the remaining ones (Burton and Lines, 1997). There are several approaches, data-driven or model-driven, currently available to predict the interbed multiples. However, they require an accurate identification of the multiple generators (Lesnikov and Owus, 2011). The identification of the origin of these multiples seems to be the most effective solutions to remove them, however it is not an easy task. The allure of Zero Offset Vertical Seismic Profiles (ZOVSPs) in having the receivers placed close to the subsurface horizons, allow both upgoing and downgoing wavefields to be recordable and separable. It's the combination of short window and long window deconvolution operators which are derived based on our knowledge of downgoing wavefield which help us to determine the multiples generators at their exact depths in the subsurface. This paper demonstrates how Zero offset VSP successfully helped to identify the major multiples generators in one of the exploratory fields in south Oman. These generators then used as an input to demultiple technique named as Extended Interbed Multiple Prediction (XIMP) that eliminates the multiples within surface seismic. As the result of the multiple elimination, the seismic to well tie tremendously improved and the reliability of the overall horizon interpretation is enhanced.

Development and evaluation of bidirectional LSTM freeway traffic forecasting models using simulation data

Long short-term memory (LSTM) models provide high predictive performance through their ability to recognize longer sequences of time series data. More recently, bidirectional deep learning models (BiLSTM) have extended the LSTM capabilities by training the input data twice in forward and backward directions. In this paper, BiLSTM short term traffic forecasting models have been developed and evaluated using data from a calibrated micro-simulation model for a congested freeway in Melbourne, Australia. The simulation model was extensively calibrated and validated to a high degree of accuracy using field data collected from 55 detectors on the freeway. The base year simulation model was then used to generate loop detector data including speed, flow and occupancy which were used to develop and compare a number of LSTM models for short-term traffic prediction up to 60 min into the future. The modelling results showed that BiLSTM outperformed other predictive models for multiple prediction horizons for base year conditions. The simulation model was then adapted for future year scenarios where the traffic demand was increased by 25–100 percent to reflect potential future increases in traffic demands. The results showed superior performance of BiLSTM for multiple prediction horizons for all traffic variables.

Impacts of model resolution on multiple prediction

Despite theoretical advancements in multiple identification and elimination, the application and success of these advancements is questionable in some cases and is limited to marine environments, especially deep water. In land seismic, however, a clear understanding of the internal multiple generators is not readily available and thus efforts to attenuate them are often quite ineffective. In this paper, we analyze, in the case of many thin layers, how the primaries and multiples are affected by fine-scale variations in the velocity model. Cross-coherence is used to measure the similarity of the original primaries/multiples and the ones generated from upscaled velocity models. The combined use of kurtosis and the cross-coherence method enables us to precisely quantify the impact of fine layering on multiples. Test results demonstrate that both surface-related multiples and internal multiples are much more sensitive than the primaries to thickness variations of the velocity model. As the thickness of each upscaled layer varies from 1m to 21m, multiples change rapidly, while primaries are almost the same. The high sensitivity of internal multiples on fine layering suggests that the detailed model information should be considered in model parameterization in the internal multiple removal, especially for model-driven methods.

Developing a Composite Measure to Represent Information Flows in Networks: Evidence from a Stock Market

This paper employs a design science approach and proposes a new composite metric, eigen attention centrality (EAC), as a proxy for information flows associated with a node that considers both attention to a node and coattention with other nodes in a network. We apply the EAC metric in the context of a financial market where nodes are individual stocks and edges are based on coattention relationships among stocks. Composite information from different channels is used to measure attention and coattention. We evaluate the effectiveness of the EAC metric on predicting abnormal returns of stocks by (1) using multiple prediction methods and (2) comparing EAC with a set of alternative network metrics. Our analysis shows that EAC significantly outperforms alternative models in predicting the direction and magnitude of abnormal returns of stocks. Using the EAC metric, we derive a stock portfolio and develop a trading strategy that provides significant and positive excess returns. Lastly, we find that composite information has significantly better predictive performance than separate information sources, and such superior performance owes to information from social media instead of traditional media.

Metamaterial Reverse Multiple Prediction Method Based on Deep Learning

Metamaterials and their related research have had a profound impact on many fields, including optics, but designing metamaterial structures on demand is still a challenging task. In recent years, deep learning has been widely used to guide the design of metamaterials, and has achieved outstanding performance. In this work, a metamaterial structure reverse multiple prediction method based on semisupervised learning was proposed, named the partially Conditional Generative Adversarial Network (pCGAN). It could reversely predict multiple sets of metamaterial structures that can meet the needs by inputting the required target spectrum. This model could reach a mean average error (MAE) of 0.03 and showed good generality. Compared with the previous metamaterial design methods, this method could realize reverse design and multiple design at the same time, which opens up a new method for the design of new metamaterials.

SL-Cloud: A Computational Resource to Support Synthetic Lethal Interaction Discovery

Synthetic lethal interactions (SLIs), genetic interactions whereby the simultaneous inactivation of two genes leads to a lethal phenotype, are promising targets for therapeutic intervention in cancer. We present SL-Cloud, an integrated resource and framework to facilitate prediction of context-specific synthetic lethal interactions using cloud-based technologies. This resource addresses two main challenges related to SLI inference, namely, the need to wrangle and preprocess large multi-omic datasets and the ability to integrate multiple prediction approaches, each of which comes with its own assumptions. We demonstrate the utility of this resource by using a set of DNA damage repair genes as the basis for predicting potential synthetic lethal interaction partners using multiple computational strategies. Context specific SLI potential can also be studied using the framework. The SL-Cloud computational resource demonstrates a variety of use cases and demonstrates the utility of this approach for customizable and extensible in silico inference of SLIs.

Development and Evaluation of Bidirectional LSTM Freeway Traffic Forecasting Models Using Simulation Data

Long short-term memory (LSTM) models provide high predictive performance through their ability to recognize longer sequences of time series data. More recently, bidirectional deep learning models (BiLSTM) have extended the LSTM capabilities by training the input data twice in forward and backward directions. In this paper, BiLSTM short term traffic forecasting models have been developed and evaluated using data from a calibrated micro-simulation model for a congested freeway in Melbourne, Australia. The simulation model was extensively calibrated and validated to a high degree of accuracy using field data collected from 55 detectors on the freeway. The base year simulation model was then used to generate loop detector data including speed, flow and occupancy which were used to develop and compare a number of LSTM models for short-term traffic prediction up to 60 minutes into the future. The modelling results showed that BiLSTM outperformed other predictive models for multiple prediction horizons for base year conditions. The simulation model was then adapted for future year scenarios where the traffic demand was increased by 25-100 percent to reflect potential future increases in traffic demands. The results showed superior performance of BiLSTM for multiple prediction horizons for all traffic variables.

A review of OBN processing: challenges and solutions

In the last decade, a significant shift in the marine seismic acquisition business has been made where ocean bottom nodes gained a substantial market share from streamer cable configurations. Ocean bottom node acquisition (OBN) can acquire wide azimuth seismic data over geographical areas with challenging deep and shallow bathymetries and complex subsurface regimes. When the water bottom is rugose and has significant elevation differences, OBN data processing faces a number of challenges, such as denoising of the vertical geophone, accurate wavefield separation, redatuming the sparse receiver nodes from ocean bottom to sea level and multiple attenuation. In this work, we review a number of challenges using real OBN data illustrations. We demonstrate corresponding solutions using processing workflows comprising denoising the vertical geophones by using all four recorded nodal components, cross-ghosting the data or using direct wave to design calibration filters for up- and down-going wavefield separation, performing one-dimensional reversible redatuming for stacking QC and multiple prediction, and designing cascaded model and data-driven multiple elimination applications. The optimum combination of the mentioned technologies produced cleaner and high-resolution migration images mitigating the risk of false interpretations.

Internal multiple prediction using inverse-scattering series with sparsity promotion — Part 1: Background, analysis, and synthetic modeling

The presence of internal multiples in seismic data can lead to artefacts in subsurface images ob-tained by conventional migration algorithms. This problem can be ameliorated by removing themultiples prior to migration, if they can be reliably estimated. Recent developments have renewedinterest in the plane wave domain formulations of the inverse scattering series (ISS) internal multipleprediction algorithms. We build on this by considering sparsity promoting plane wave transformsto minimize artefacts and in general improve the prediction output. Furthermore, we argue forthe usage of demigration procedures to enable multidimensional internal multiple prediction withmigrated images, which also facilitate compliance with the strict data completeness requirementsof the ISS algorithm. We believe that a combination of these two techniques, sparsity promotingtransforms and demigration, pave the way for a wider application to new and legacy datasets.