nasa-petal/PeTaL-labeller
The PeTaL labeler labels journal articles with biomimicry functions.
Jupyter NotebookUnlicense
Issues
- 0
Preprocess_golden.py
#88 opened by pjuangph - 1
Get to 90% precision
#22 opened by bruffridge - 0
- 1
Change output of labeller from absolute "selection" to "ranking" using confidence scores.
#83 opened by bruffridge - 1
Match vocab file PeTaL.emb
#86 opened by pjuangph - 2
Use open-source Snorkel to create labelling functions to expand our training dataset.
#65 opened by bruffridge - 0
Look into Label Studio to help increase the size of our labelled dataset for training
#80 opened by bruffridge - 0
- 0
- 0
Use a large ensemble of single-label classifiers (so, treat all of the labels independently, ignore the hierarchy, and we have a hundred separate yes/no tasks) and see if this works better than MATCH
#74 opened by bruffridge - 0
- 1
Compare MATCH results to auto-sklearn
#56 opened by bruffridge - 1
Try using a tree of multilabel classifiers
#81 opened by bruffridge - 4
Run Match on just level 1 labels
#73 opened by bruffridge - 4
See how replacing random weights with pretrained and fine-tuned weights in MATCH affects performance
#72 opened by bruffridge - 5
Produce metrics to show which labels are being classified correctly and which aren't, and how they're being misclassified.
#61 opened by bruffridge - 1
Add a description of metrics to MATCH's README
#77 opened by bruffridge - 1
Try Google Cloud AutoML Natural Language's multilabel text classification on golden dataset
#78 opened by bruffridge - 0
try to use a language model like GPT-2 for its general-purpose language understanding capabilities (and then integrate it with the MATCH classification task somehow)
#76 opened by bruffridge - 5
- 1
Analyze our training dataset to discover ways to improve it to improve labelling accuracy
#52 opened by bruffridge - 1
- 1
- 6
Look into using MATCH to improve the labeler
#42 opened by bruffridge - 3
- 1
rerun ablation study
#68 opened by bruffridge - 0
- 1
Replace MAG topics with another topic taxonomy
#59 opened by bruffridge - 1
See if we can expand our training dataset by leveraging NLM MeSH labels or Microsoft Academic topics
#21 opened by bruffridge - 0
Create a POC for integrating a Colab classification model with weights and biases
#44 opened by bruffridge - 1
Do k-fold cross validation to generate ablation study results for including MAG and MeSH labels.
#58 opened by bruffridge - 1
Plot a graph that shows how much adding additional training data improves labeller accuracy.
#40 opened by bruffridge - 1
- 0
Create a label hierarchy for MATCH input
#47 opened by bruffridge - 3
- 0
Prepare train/test data for MATCH
#43 opened by bruffridge - 1
CORE dataset vs. Semantic Scholar
#17 opened by bruffridge - 4
- 1
- 0
Identify the characteristics of a paper describing thermal management in nature
#39 opened by bruffridge - 2
Test out how well the ML model abstains from applying labels to abstracts that don't belong within the biomimicry taxonomy.
#19 opened by bruffridge - 4
duplicate code?
#31 opened by bruffridge - 0
Investigate TF-IDF
#36 opened by pjuangph - 1
Adapt Huggingface predictor to PubMed
#28 opened by pjuangph - 0
Add code to shruti's labeller to estimate accuracy of train and validation datasets
#27 opened by pjuangph - 0
- 0
- 1
- 0
Look into domain adaptive pre-training
#24 opened by bruffridge - 0
Instrument Labebller using WandB
#18 opened by pyvelepor