RDFSCompletionReasoningLSTM issueshttps://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues2021-10-07T16:06:23-05:00https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/12Potential Cross Validation Problem2021-10-07T16:06:23-05:00bpankaskPotential Cross Validation ProblemI think the current cross validation method just repeats the same train/validation split k times instead of shuffling these?I think the current cross validation method just repeats the same train/validation split k times instead of shuffling these?bpankaskbpankaskhttps://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/11Testing Diff Data2021-10-07T16:32:03-05:00bpankaskTesting Diff DataNeed to add functionality to test on different data then trained.Need to add functionality to test on different data then trained.bpankaskbpankaskhttps://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/10Update Tests for main.py2021-10-06T18:31:27-05:00bpankaskUpdate Tests for main.pyTests are outdated and need to be updated. If there is a way to write them so that they work with synthetic examples this would be ideal.Tests are outdated and need to be updated. If there is a way to write them so that they work with synthetic examples this would be ideal.bpankaskbpankaskhttps://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/4Random Prediction Generator2021-07-27T08:34:45-05:00bpankaskRandom Prediction Generator*Created by: bpankask*
Need to add capability to guess random predictions to test smart predictions against.*Created by: bpankask*
Need to add capability to guess random predictions to test smart predictions against.https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/5Random Predictions 2021-07-30T10:10:08-05:00bpankaskRandom Predictions *Created by: bpankask*
Doesn't ever seem to be guessing correctly. Need to figure out what Aaron did which at least gave him a baseline. *Created by: bpankask*
Doesn't ever seem to be guessing correctly. Need to figure out what Aaron did which at least gave him a baseline. https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/9Local Label Issue2021-09-22T15:27:25-05:00bpankaskLocal Label Issue*Created by: bpankask*
Original project reassigns label numbers for each test samples. This limits the number of labels and therefore the number of guesses the LSTM can make. This should fix the problem I had where the LSTM was never ...*Created by: bpankask*
Original project reassigns label numbers for each test samples. This limits the number of labels and therefore the number of guesses the LSTM can make. This should fix the problem I had where the LSTM was never guessing anything correctly. It should also be noted that the LSTM was getting close so a deeper model would probably get some correct if we didn't want to do local labels. https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/2Noisy Data2021-07-14T18:33:32-05:00bpankaskNoisy Data*Created by: bpankask*
Need to add functionality which will generate, train, and evaluate data which contains bad triples.*Created by: bpankask*
Need to add functionality which will generate, train, and evaluate data which contains bad triples.https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/6String IRI and Label Evaluation Problem2021-08-16T09:35:17-05:00bpankaskString IRI and Label Evaluation Problem*Created by: bpankask*
Both the IRI and label true/false positives/negatives are always going to get the same answer because they are based off of each other. To get the IRI you need the label. *Created by: bpankask*
Both the IRI and label true/false positives/negatives are always going to get the same answer because they are based off of each other. To get the IRI you need the label. https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/7Fix Final Average Error2021-07-27T08:32:46-05:00bpankaskFix Final Average Error*Created by: bpankask*
Final average isn't logging.*Created by: bpankask*
Final average isn't logging.https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/8Loss of KB Sample After Splitting Data2021-08-04T08:17:33-05:00bpankaskLoss of KB Sample After Splitting Data*Created by: bpankask*
Loss of some KB samples after splitting the data for cross validation.*Created by: bpankask*
Loss of some KB samples after splitting the data for cross validation.https://gitlab.cs.ksu.edu/bpankask/RDFSCompletionReasoningLSTM/-/issues/1Bad Triple Capabilities 2021-07-14T18:33:53-05:00bpankaskBad Triple Capabilities *Created by: bpankask*
Need to eventually add functionality to allow for the generation, training, and evaluation of bad triples which will add noise.*Created by: bpankask*
Need to eventually add functionality to allow for the generation, training, and evaluation of bad triples which will add noise.