Open source the implementation of Graph Agreement Models.

PiperOrigin-RevId: 266445340

Author: Neural-Link Team

neural_structured_learning/research/README.md

@@ -1,8 +1,17 @@
 # Research projects in Neural Structured Learning
 
-The implementations of A2N [1] are provided in the "a2n" folder on a strict "as
+## A2N
+
+The implementations of A2N [1] are provided in the `a2n` folder on a strict "as
 is" basis, without warranties or conditions of any kind. Also, these
 implementations may not be compatible with certain TensorFlow versions (such as
 2.0 or above) or Python versions.
 
 [[1] T. Bansal, D. Juan, S. Ravi and A. McCallum. "A2N: Attending to Neighbors for Knowledge Graph Inference." ACL 2019](https://www.aclweb.org/anthology/P19-1431)
+
+## Graph Agreement Models
+
+The implementations of Graph Agreement Models (GAMs) are provided in the `gam`
+folder on a strict "as is" basis, without warranties or conditions of any kind.
+Also, these implementations may not be compatible with certain TensorFlow
+versions (such as 2.0 or above) or Python versions.

neural_structured_learning/research/gam/README.md

@@ -0,0 +1,39 @@
+Neural structured learning methods such as Neural Graph Machines [1], Graph
+Convolutional Networks [2] and their variants have successfully combined the
+expressiveness of neural networks with graph structures to improve on learning
+tasks. Graph Agreement Models is a technique that can be applied to these
+methods to handle the noisy nature of real world graphs. Traditional graph-based
+algorithms, such as label propagation, were designed with the underlying
+assumption that the label of a node can be imputed from that of the neighboring
+nodes and edge weights. However, most real world graphs are either noisy or have
+edges that do not correspond to label agreement uniformly across the graph.
+Graph Agreement Models(GAM) introduce an auxiliary model that predicts the
+probability of two nodes sharing the same label as a learned function of their
+features. This agreement model is then used when training a node classification
+model by encouraging agreement only for those pairs of nodes that it deems
+likely to have the same label, thus guiding its parameters to better local
+optima. The classification and agreement models are trained jointly in a
+co-training fashion.
+
+This code repository contains an implementation of Graph Agreement Models. The
+code is organized into the following folders:
+
+*   data: Classes and methods for accessing semi-supervised learning datasets.
+*   models: Classes and methods for classification models and graph agreement
+    models.
+*   trainer: Classes and methods for training the classification models, and
+    agreement models individually as well as in a co-traioning fashion.
+*   experiments: Python run script for training Graph Agreement Models on MNIST
+    and other datasets.
+
+The implementations of Graph Agreement Models (GAMs) are provided in the `gam`
+folder on a strict "as is" basis, without warranties or conditions of any kind.
+Also, these implementations may not be compatible with certain TensorFlow
+versions (such as 2.0 or above) or Python versions.
+
+## Reference
+
+[[1] T. Bui, S. Ravi and V. Ramavajjala. "Neural Graph Learning: Training Neural Networks Using Graphs." WSDM 2018](https://ai.google/research/pubs/pub46568.pdf)
+
+
+[[2] T. Kipf and M. Welling. "Semi-supervised classification with graph convolutional networks." ICLR 2017](https://arxiv.org/pdf/1609.02907.pdf)

neural_structured_learning/research/gam/__init__.py

@@ -0,0 +1,13 @@
+# Copyright 2019 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.

neural_structured_learning/research/gam/data/__init__.py

@@ -0,0 +1,13 @@
+# Copyright 2019 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.