Skip to content
jerryuhoo edited this page Nov 7, 2019 · 4 revisions

Lack of experience with technologies

Description: No one in the team is familiar with machine learning leading to a potentially difficult time optimizing the current program

Risk Owner: Plant Phenotyping Team

Assigned To: Plant Phenotyping Team

Status: At risk

Risk Indicator: Programming

Risk Indicator Description: Try to write a new function but do not know how to optimize machine learning algorithm

Probability (percentage): 100%

Avoidance Plan: No plan because we have to solve it directly rather than avoidance.

Risk Impact (severity): High

Risk Impact Description: It stops the progress of programming

Mitigation Plan: Learn machine learning ahead of time

Contingency/Recovery Plan: Ask the coach for help.

HCC Server could be down

Description: Maintenance or unexpected error could cause the HCC Server to go down delaying work time

Risk Owner: HCC

Assigned To: Plant Phenotyping Team

Status: At risk

Risk Indicator: Cannot access server

Risk Indicator Description: Logging into the server via crane or etc. will not allow user to enter any commands

Probability (percentage): 10%

Avoidance Plan: Avoid actions or commands that might harm the server

Risk Impact (severity): High

Risk Impact Description: It stops developing and testing on the server

Mitigation Plan: Work locally or on other tasks

Contingency/Recovery Plan: Wait patiently for the server to be fixed

Lack of documentation

Description: Since we are already given a codebase, lack of documentation could hurt our speed and accuracy of learning the code

Risk Owner: Schnablelab developers

Assigned To: Phillip Nguyen

Status: At risk

Risk Indicator: Do not understand many functions/methods

Risk Indicator Description: Current code lacks enough comments or supporting documents that would allow us to understand the code

Probability (percentage): 20%

Avoidance Plan: Cannot really be avoided

Risk Impact (severity): Moderate

Risk Impact Description: If this happens, we will be slowed down due to lack of knowledge but not stopped

Mitigation Plan: Ask for more documentation and/or learn on our own

Contingency/Recovery Plan: Learn on our own or look online

Unable to increase code performance

Description: Code may be optimized already or we lack the tools or knowledge to either find problems with code performance or optimize code

Risk Owner: Plant Phenotyping Team

Assigned To: Zhenghui Su

Status: At risk

Risk Indicator: Performance does not increase

Risk Indicator Description: We were unable to locate the slowdowns or our changes show no signs of significant performance improvements

Probability (percentage): 15%

Avoidance Plan: Study hard about machine learning and ask for as much information as possible about profiling and optimization tips from experts

Risk Impact (severity): High

Risk Impact Description: One of the main purposes of this project is to optimize the performance of Schnablelab. Failing to do this would make this project lose a lot of value.

Mitigation Plan: Learn all we can about the current performance of Schnablelab and study optimization techniques

Contingency/Recovery Plan: Ask for help from experts or find a new approach

Unable to store/retrieve information directly to/from GLOBUS - Integration Obstacles

Description: If we are not able to take the input directly from the source and return it to the source location it will create data duplicates. Since the data is very large and the Schnable Lab already is having issues with data storage, doubling the data is not possible.

Risk Owner: Plant Phenotyping Team

Assigned To: Plant Phenotyping Team

Status: At Risk

Risk Indicator: Current functionality requires locally stored files

Risk Indicator Description: We currently are unaware of ways to retrieve and store data directly in GLOBUS for our current working project. We aren’t even aware if we have the required permissions to do so. This means we have data copies we are using for testing locally.

Probability (percentage): 25%

Avoidance Plan: Find a way to interact directly with those locations for a specified data set to ensure not data replication during the workflow.

Risk Impact (severity): Very High

Risk Impact Description: If our solution is unable to be performed without data replication then it is essentially useless.

Mitigation Plan: Spend a significant amount of time looking into drawing the information to/from the GLOBUS servers. Verify if permissions are needed, the output and input is not creating waste, and that our current solution is viable.

Contingency/Recovery Plan: Contact our sponsors about alternative ways to access the data and find new methods for the desired functionality.

Unable to generate accurate plant data - Lack of Experience with Technologies

Description: Due to our unfamiliarity with opencv image processing and TensorFlow machine learning libraries, as well as, a test data set which is not highly accurate we may not be able to truly produce high accuracy output data. We need to apply these things to our predicted images to help find plant height, height to top leaf, number of leaves, and the dimensions of the plants’ inflorescence.

Risk Owner: Plant Phenotyping Team

Assigned To: Plant Phenotyping Team

Status: At Risk

Risk Indicator: General lack of knowledge with the technologies for simply retrieving some data and zero experience verifying the newly generated data accuracy, especially for scientific and research purposes.

Risk Indicator Description: Our team is currently uncertain of the means of producing the desired data by the Schnable Lab and it will take some time to learn how to do so. Even assuming we generate reasonable data, we have no current understanding on the data accuracy verification of results.

Probability (percentage): 70%

Avoidance Plan: Go into our development for this new functionality with a focus on accuracy and not simply results.

Risk Impact (severity): Moderate

Risk Impact Description: If we are able to get results for the desired data and it is unusable for scientific research, we will have to redo our previous work and retrain the machine learning algorithm we created to collect the data for higher accuracy. This could mean that we might have to create quite a large amount of test data for further data accuracy in training.

Mitigation Plan: Try to ensure that we are accurately producing results. We may need to apply statistical analysis to the data for understanding its level of accuracy.

Contingency/Recovery Plan: Work on developing more accurate training data so that all results have a higher level of accuracy.