Siamese networks
18 likes10,505 views
The document discusses Siamese networks and their application in textual record matching, highlighting their capability to improve similarity measures while addressing issues like model bias and unexplainable results. It outlines their architecture, training datasets, loss functions, and potential use cases such as natural language processing and address matching, while also detailing the advantages and disadvantages of these networks. The conclusion notes that Siamese networks leverage relationships between data points and can generalize to unseen data, making them useful for tasks where labeling similar and dissimilar points is feasible.
![Training Dataset
● Siamese Networks must be trained on data that has two
inputs and a target similarity.
– [‘input a1’, ‘input a2’, 1]
– [‘input a2’, ‘input a3’, 1]
– [‘input a2’, ‘input b1’, -1]
– …
● There must be similar inputs (+1) and dissimilar inputs (-1).
● Most studies have shown that the ratio of dissimilar to
similar is optimal around:
– Between 2:1 and 10:1.
– This depends on the problem and specificity of the model needed.](https://image.slidesharecdn.com/siamesenetworks-170626141840/85/Siamese-networks-6-320.jpg)
Siamese networks
- 1. Siamese Networks Textual Record Matching By Nick McClure @nfmcclure
- 2. Outline ● Motivation – Why are siamese networks useful? – Why should we care? – What are the benefits? ● Structure and Loss – What are they? – How does the loss work? ● Use Cases ● Address Matching Example Code
- 3. Motivation ● Neural Networks can have unintentional behaviors. – Outliers – Model Bias – Unexplainable results ● Siamese Networks impose a structure that helps combat these problems. ● Siamese Networks allow us to use more data points than we would have in other common cases. – Allow us to use relationships between data points! Blue Red
- 4. Structural Definition ● Siamese networks train a similarity measure between labeled points. ● Two input data points (textual embeddings, images, etc…) are run simultaneously through a neural network and are both mapped to a vector of shape Nx1. ● Then a standard numerical function can measure the distance between the vectors (e.g. the cosine distance).
- 5. Structural Definition Input A Input B Neural Network Architecture Neural Network Architecture Vector A Output Vector B Output Cosine Similarity -1<=output<=1 Same Parameters Same Structure
- 6. Training Dataset ● Siamese Networks must be trained on data that has two inputs and a target similarity. – [‘input a1’, ‘input a2’, 1] – [‘input a2’, ‘input a3’, 1] – [‘input a2’, ‘input b1’, -1] – … ● There must be similar inputs (+1) and dissimilar inputs (-1). ● Most studies have shown that the ratio of dissimilar to similar is optimal around: – Between 2:1 and 10:1. – This depends on the problem and specificity of the model needed.
- 7. Training Dataset ● Since we have to generate similar and dissimilar pairs, the actual amount of training data is quite higher than normal. ● For example, in the UCI machine learning data set of spam/ham text messages, there are 656 observations, 577 are ham and only 79 are spam. ● With the siamese architecture, we can consider up to 79 choose 2 = 3,081 similar spam comparisons and 577 choose 2 = 166,176 similar ham comparisons, while having 45,583 total dissimilar comparisons!!!
- 8. Dealing with New Data ● Another benefit is that siamese similarity networks generalize to inputs and outputs that have never been seen before. – This makes sense when comparing to how a person can make predictions on unseen instances and events.
- 9. Help Explain Results ● With siamese networks, we can always list the nearest points in the output-vector space. Because of this, we can say that a data point has a specific label because it is nearest to a set of points. ● This type of explanation does not depend on how complicated the internal structure is.
- 10. Siamese Loss Function: Contrastive Loss ● The loss function is a combination of a similar-loss (L+) and dissimilar-loss (L-).
- 11. How Does the Backpropagation Work? ● Siamese networks are constrained to have the same parameters in both sides of the network. ● To train the network, a pair of output vectors needs to be either closer together (similar) or further apart (dissimilar). ● It is standard to average the gradients of the two ‘sides’ before performing the gradient update step.
- 12. Potential Use Cases ● Natural Language Processing: – Ontology creation: How similar are words/phrases? – Job Title Matching: ‘VP of HR’ == ‘V.P. of People’ – Topic matching: Which topic is this phrase referring to? ● Others: – Image recognition – Image search – Signature/Speech recognition
- 13. Address Matching! ● Input addresses can have typos. We need to be able to process these addresses and match them to a best address from a canonical truth set. ● E.g. ‘123 MaiinSt’ matches to ‘123 Main St’. – Why? Fat fingers, image→text translation errors, encoding errors, etc... ● Our siamese network will be a bidirectional LSTM with a fully connected layer on the top.
- 14. Address Matching! TensorFlow Demo! Here: https://github.com/nfmcclure/tensorflow_cookbook Navigate to Chapter 09: RNNs, then Section 06: Training a Siamese Similarity Measure
- 15. Conclusions and Summary ● Advantages: – Can predict out-of-training-set data. – Makes use of relationships, using more data. – Explainable results, regardless of network complexity. ● Disadvantages: – More computationally intensive (precomputation helps however). – More hyperparameters and fine-tuning necessary. – Generally, more training needed. ● When to use: – Want to exploit relationships between data points. – Can easily label ‘similar’ and ‘dissimilar’ points. ● When not to use:
- 16. Further References ● Signature Verification with fully connected siamese newtworks, 1995, Yann LeCun, et. al., Bell Labs, http://papers.nips.cc/paper/769-signature-verification-using-a-siamese-time- delay-neural-network.pdf ● Attention based CNN for sentence similarity, 2015, https://arxiv.org/pdf/1512.05193v2.pdf ● Learning text similarities with siamese RNNs, 2016, http://anthology.aclweb.org/W16-1617 ● Sketch-based Image Retrieval via Siamese CNNs, 2016, http://qugank.github.io/papers/ICIP16.pdf