A Concept of Ostracods Project Proposal
*There are many articles describing the morphology of ostracods, both in words and images. Some fossils are well preserved and some are partially damaged but still recognizable. This would provide a spectrum on the quality of ostracods’ image adequate for the training samples
*There are also many articles describing ostracods of various geologic ages, ranging from Pleistocene to Silurian (?). This gives another dimension of information for the training.
*One goal of this study is to construct an ostracod database that collects as many currently available ostracod images and related information as possible. Then, the database will be used to train a machine learning module.
The trained intelligent module will be tested by a set of well-studied ostracods and then be used to identify an unknown set of ostracods in the cyclothem of coal bed in the Northern China Basin.
A few amateur questions about the machine learning system:
*What is available to start the construction
*How many layers neural network
*Two-way (?) training?
*How to adjust and fine-tune?
*Computing power needed?
Basic project: The AI is able to successfully pick up ostracods among various microfossils.
Standard project: The AI is not able to identify ostracod, but tell the species and age of ostracod.
Good project: The AI is able to identify well preserved and partially preserved ostracods.
Outstanding project: The AI is able to identify ostracod (species, age) in a thin section or a rock slab/section. (e.g. drill core, outcrop)
Excellent project: The AI is able to identify ill-preserved ostracods in a thin section or a rock slab/section.
To Build an Artificial Intelligent System
for the Identification of Ostracods
建立一个介型虫鑑定的人工智能系统