The need for artificial intelligence (AI) and machine learning (ML) models to optimize data center (DC) operations increases as the volume of operations management data upsurges tremendously. These strategies can assist operators in better understanding their DC operations and help them make informed decisions upfront to maintain service reliability and availability. The strategies include developing models that optimize energy efficiency, identifying inefficient resource utilization and scheduling policies, and predicting outages. In addition to model hyperparameter tuning, feature subset selection (FSS) is critical for identifying relevant features for effectively modeling DC operations to provide insight into the data, optimize model performance, and reduce computational expenses. Hence, this paper introduces the Shapley Additive exPlanation (SHAP) values method, a class of additive feature attribution values for identifying relevant features that is rarely discussed in the literature. We compared its effectiveness with several commonly used, importance-based feature selection methods. The methods were tested on real DC operations data streams obtained from the ENEA CRESCO6 cluster with 20,832 cores. To demonstrate the effectiveness of SHAP compared to other methods, we selected the top ten most important features from each method, retrained the predictive models, and evaluated their performance using the MAE, RMSE, and MPAE evaluation criteria. The results presented in this paper demonstrate that the predictive models trained using features selected with the SHAP-assisted method performed well, with a lower error and a reasonable execution time compared to other methods.
A rock-physics study and AVO modeling study have been completed to assist in the interpretation of seismic amplitude and AVO anomalies in the Shipwreck Trough of the offshore Otway Basin of southeastern Australia. Elastic log data, core data (both full and sidewall), and associated thin-section analysis of composition and texture were available on several wells, and these data are important in calibrating proposed rock-physics models that suggest that incorporating cement is critical to understanding anomalies in seismic and measured log data. Lithoprobability volumes based on conventional interpretation paradigms, such as low VP/VS values indicating gas presence, that do not incorporate an understanding of rock physics lead to biased interpretations. Ratios in particular can be misleading because there is ambiguity about whether an anomalous ratio is driven by the numerator or denominator. As a classic gas indicator, low VP/VS values are interpreted to be driven by a decrease in VP associated with gas replacing brine in a rock. Using Lamé impedance terms λρ and μρ, however, provides an alternative interpretation template that does not use ratios and can improve insight into rock properties. As in this recent case study, using lambda-mu-rho (LMR) can be an important tool when shear velocity has increased relative to compressional velocity, irrespective of any pore-fluid change. In the reservoir rocks of the Shipwreck Trough, low VP/VS in both gas-saturated and brine-filled sandstone is caused by quartz cement. This presents a substantial challenge to the use of a standard rock-physics template. In LMR space, however, low VP/VS data points clearly are characterized by high shear rigidity — an important point to recognize and incorporate into AVO interpretation workflows.
Rugose seabed and near seafloor features in marine seismic datasets such as canyons, mass transport deposits and paleo-channels, present a major imaging challenge that can be overcome by a rigorous and careful approach in the earth model building process. To obtain an accurate image of the deeper target levels, a modelling and data driven update of the distinctive velocity characteristics of the overburden features are necessary. Without adequately addressing the complexity of the shallow velocity field, the final depth image of the target intervals can be poorly focussed and contaminated with non-geologic structural distortions. Inadequate corrections ultimately have an adverse impact upon the interpretation of the dataset.
This paper presents a successful earth modelling approach used to obtain an accurate depth image for a marine dataset located on the shelf break in the Otway Basin. The case study area includes extensive seafloor canyons and associated paleo-channels, requiring the strategic use of several geologically constrained model updating technologies in order to obtain a final imaged section free of velocity related structural distortions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.