A transparent Open-Box learning network provides insight to complex systems and a performance benchmark for more-opaque machine learning algorithms
Journal Title: Advances in Geo-Energy Research - Year 2018, Vol 2, Issue 2
Abstract
It is now commonplace to deploy neural networks and machine-learning algorithms to provide predictions derived from complex systems with multiple underlying variables. This is particularly useful where direct measurements for the key variables are limited in number and/or difficult to obtain. There are many petroleum systems that fit this description. Whereas artificial intelligence algorithms offer effective solutions to predicting these difficult-to-measure dependent variables they often fail to provide insight to the underlying systems and the relationships between the variables used to derive their predictions are obscure. To the user such systems often represent “black boxe”. The novel transparent open box (TOB) learning network algorithm described here overcomes these limitations by clearly revealing its intermediate calculations and the weightings applied to its independent variables in deriving its predictions. The steps involved in building and optimizing the TOB network are described in detail. For small to mid-sized datasets the TOB network can be deployed using spreadsheet formulas and standard optimizers; for larger systems coded algorithms and customised optimizers are easy to configure. Hybrid applications combining spreadsheet benefits (e.g. Microsoft Excel Solver) with algorithm code are also effective. The TOB learning network is applied to three petroleum datasets and demonstrates both its learning capabilities and the insight to the modelled systems that it is able to provide. TOB is not proposed as a replacement for neural networks and machine learning algorithms, but as a complementary tool; it can serve as a performance benchmark for some of the less transparent algorithms.
Authors and Affiliations
David A. Wood
Keywords
Related Articles
- EP ID EP427917
- DOI 10.26804/ager.2018.02.04
- Views 61
- Downloads 0
A transparent Open-Box learning network provides insight to complex systems and a performance benchmark for more-opaque machine learning algorithms
It is now commonplace to deploy neural networks and machine-learning algorithms to provide predictions derived from complex systems with multiple underlying variables. This is particularly useful where direct measurement...
Quantitative characterization of micropore structure for organic-rich Lower Silurian shale in the Upper Yangtze Platform, South China: Implications for shale gas adsorption capacity
The pores in shales are mainly of nanometer-scale, and their pore size distribution is very important for shale gas storage and adsorption capacity, especially micropores having widths less than 2 nm, which contribute to...
Insights on the gas permeability change in porous shale
Due to abundant nanoscale pores developed in shale, gas flow in shale presents a complex dynamic process. This paper summarized the effects from effective stress increase, shale matrix shrinkage, gas slippage and Knudsen...
Discussion on seepage governing equations for low permeability reservoirs with a threshold pressure gradient
Low permeability reservoirs account for an increasing proportion of oil production. Threshold pressure gradient is an important factor that governs the flow in low permeability porous media. The 1-D seepage governing equ...
A comprehensive review of pore scale modeling methodologies for multiphase flow in porous media
Multiphase flow in porous media is relevant to amount of engineering processes, such as hydrocarbon extraction from reservoir rock, water contamination, CO2 geological storage and sequestration. Pore scale modeling, as a...