This folder contains an implementation for Abductive Learning (ABL), which is built from ABLkit, an efficient Python toolkit for ABL.

ABL is a novel paradigm that integrates machine learning and logical reasoning in a unified framework. It is suitable for tasks where both data and (logical) domain knowledge are available.

ABLkit is an efficient Python toolkit for ABL, which encapsulates advanced ABL techniques, providing users with an efficient and convenient toolkit to develop dual-driven ABL systems that leverage the power of both data and knowledge.

If the use of a Prolog-based knowledge base is necessary, please install SWI-Prolog:

For Linux users:

sudo apt-get install swi-prolog

For Windows and Mac users, please refer to the SWI-Prolog Install Guide.

ABLkit requires user data to be either structured as tuple (X, gt_pseudo_label, Y) or ListData object with X, gt_pseudo_label and Y attributes.

X is a list of input examples containing instances, gt_pseudo_label is the ground-truth label of each example in X and Y is the ground-truth reasoning result of each example in X. Note that gt_pseudo_label is only used to evaluate the machine learning model's performance but not to train it.

For tabular data, one may use lambdaLearn.Algorithm.ABLkit.data.DataConverter to convert them to the aforementioned ABLkit data. Below shows an example:

from lambdaLearn.Dataset.Tabular.BreastCancer import BreastCancer
from lambdaLearn.Algorithm.ABLkit.data import DataConverter

# Get a lambdaLearn tabular dataset: BreastCancer
breast_dataset = BreastCancer(labeled_size=0.1, stratified=True, shuffle=True)

# Instantiate an DataConverter
data_converter = DataConverter()

# Convert BreastCancer
ABLkit_dataset = data_converter.convert_lambdalearn_to_tuple(breast_dataset)

For other types of data, please manually convert them.

ABLkit requires user to provide a base model for machine learning, and then wrap it into an instance of ABLModel. To convert LambdaLearn model to the aforementioned ABLkit learning models, one may use lambdaLearn.Algorithm.ABLkit.learning.ModelConverter. Below shows an example:

from torch import nn
from torch.optim import RMSprop, lr_scheduler
from lambdaLearn.Algorithm.SemiSupervised.Classification.FixMatch import FixMatch
from lambdaLearn.Network.LeNet5 import LeNet5
from lambdaLearn.Algorithm.ABLkit.learning import ModelConverter

# Get a lambdaLearn model: FixMatch
model = FixMatch(
    network=LeNet5(),
    threshold=0.95,
    lambda_u=1.0,
    mu=7,
    T=0.5,
    epoch=1,
    num_it_epoch=2**20,
    num_it_total=2**20,
    device="cuda",
)

loss_fn = nn.CrossEntropyLoss(label_smoothing=0.2)
optimizer_dict = dict(optimizer=RMSprop, lr=0.0003, alpha=0.9)
scheduler_dict = dict(scheduler=lr_scheduler.OneCycleLR, 
                      max_lr=0.0003, 
                      pct_start=0.15, 
                      total_steps=200)

# Instantiate a ModelConverter
converter = ModelConverter()

# Convert FixMatch
base_model = converter.convert_lambdalearn_to_basicnn(
    model, loss_fn=loss_fn, optimizer_dict=optimizer_dict, scheduler_dict=scheduler_dict
)

After completing the two conversions mentioned above, the construction of other components in a ABL system is the same as that of ABLkit. To explore detailed information of ABL and ABLkit, please refer to - ABLkit's document.

For more information about ABL, please refer to: Zhou, 2019 and Zhou and Huang, 2022.

@article{zhou2019abductive,
  title     = {Abductive learning: towards bridging machine learning and logical reasoning},
  author    = {Zhou, Zhi-Hua},
  journal   = {Science China Information Sciences},
  volume    = {62},
  number    = {7},
  pages     = {76101},
  year      = {2019}
}

@incollection{zhou2022abductive,
  title     = {Abductive Learning},
  author    = {Zhou, Zhi-Hua and Huang, Yu-Xuan},
  booktitle = {Neuro-Symbolic Artificial Intelligence: The State of the Art},
  editor    = {Pascal Hitzler and Md. Kamruzzaman Sarker},
  publisher = {{IOS} Press},
  pages     = {353--369},
  address   = {Amsterdam},
  year      = {2022}
}