Various models and code for paraphrase identification implemented in Tensorflow (1.1.0).
I took great care to document the code and explain what I'm doing at various steps throughout the models; hopefully it'll be didactic example code for those looking to get started with Tensorflow!
So far, this repo has implemented:
A basic Siamese LSTM baseline, loosely based on the model in Mueller, Jonas and Aditya Thyagarajan. "Siamese Recurrent Architectures for Learning Sentence Similarity." AAAI (2016).
A Siamese LSTM model with an added "matching layer", as described in Liu, Yang et al. "Learning Natural Language Inference using Bidirectional LSTM model and Inner-Attention." CoRR abs/1605.09090 (2016).
The more-or-less state of the art Bilateral Multi-Perspective Matching model from Wang, Zhiguo et al. "Bilateral Multi-Perspective Matching for Natural Language Sentences." CoRR abs/1702.03814 (2017).
PR's to add more models / optimize or patch existing ones are more than welcome! The bulk of the model code resides in duplicate_questions/models
A lot of the data processing code is taken from / inspired by allenai/deep_qa, go check them out if you like how this project is structured!
2.【代码】Face classification and detection
Real-time face detection and emotion/gender classification using fer2013/IMDB datasets with a keras CNN model and openCV.
IMDB gender classification test accuracy: 96%.
fer2013 emotion classification test accuracy: 66%.
Emotion/gender classification of the B-IT-BOTS robotics team :)
3.【博客】TensorFlow Dev Summit 2017: Integrating Keras and TensorFlow
I am briefly sharing a video from the last TensorFlow Dev Summit in February 2017. My choice has fallen to a presentation by François Chollet of the deep learning library API Keras and its integration with TensorFlow. As Dr. Chollet explains, Keras integrated with TensorFlow promises to streamline deep learning frameworks in ways that will be increasingly user-friendly, rendering the mass adoption of these software developments a more feasible reality:
4.【博客】Understanding and Implementing CycleGAN in TensorFlow
Transferring characteristics from one image to another is an exciting proposition. How cool would it be if you could take a photo and convert it into the style of Van Gogh or Picasso!
5.【资源】Deep Learning Resources