2019 May 07¶
Location: Coors Tech¶
Attending¶
Jihyun Andy Thomas Antoine Bin Iga
Key takeaways¶
- The well log GAN project to be a tutorial-like paper in Geophysics
- Well log GAN peeps will meet weekly over the summer (TBD)
- In-person meetings to resume on campus in Fall semester
Problems¶
- Data still needs to be better cleaned
- Identify other potential well data types besides resistivity
- Check the units of the well data
- Hani’s like a Pokemon, We gotta catch em.
Updated notes on well log paper¶
- Scope of paper
- Stick to CNN-based GAN
- Multi-dimensional training data after 1D success
- Tone is tutorial-like
- Motivation
- lack of geosci training data
- well logs are simple and high fidelity
- ubiquitous
- oil,water,mining
- generator
- no petro-phys model/simulator
- act as a prior for inversion
- mention gradients?
- links different types of logs
- well log data recovery
- unsupervised
- Contributions of paper
- published curated dataset
- published trained network
- GAN tutorial
- foothold into further GAN work
- Outline of paper
- Intro
- Geology of Kansas - postpone to Fall
- GAN background - Andy, Jihyun, Bin, and Iga
- RILD background - Andy
- Methods
- QC/data cleaning - Andy
- GAN - Andy, Jihyun, Bin, and Iga
- Cross-validation - Thomas
- spatial stats
- wavelet transform
- Fourier transform
- PCA
- Results
- QC/data cleaning - Andy
- GAN - Andy, Jihyun, Bin, and Iga
- Cross-validation - Thomas
- spatial stats
- wavelet transform
- Fourier transform
- PCA
- Discussion
- each result item
- Answer: What did/didn’t the network learn
- speculation on use cases
- recommendations for changes
- predictions/future work
- Intro
- Task leader breakdown
- Jihyun: Gan literature review
- Thomas: statistical analysis
- Andy: Data cleaning
- may need help
- Download zipped las files from KGS (link on mlgp website)
- check units: depth and Ohm-m
- Do our best to handle well geometry
- ID Nans
- Handle NaNs
- ID most popular aliases
- write functions in .py file
- Function 1
- takes: a well log, mneumonic
- returns mask for nan, depth
- Function 2
- takes: data, mask, depths
- returns: ‘clean’ data
- Interpolate over no more than 10 ft
Agenda for next meeting¶
- Go get a burger
- Communicate work done over the summer
- Line up potential speakers from GP department for the fall