Skip to main content

In the World of Document Similarity

How does a human infer whether two documents are similar? This question has dazzled cognitive scientists, and is one area under which a lot of research is taking place. As of  now there is no product that is able to match or surpass human capability in finding the similarity in documents. But things are improving in this domain, and companies such as IBM and Microsoft are investing a lot in this area.

We at Cere Labs, an Artificial Intelligence startup based in Mumbai, also are working in this area, and have applied LDA and Word2Vec techniques, both giving us promising results:

Latent Dirichlet Allocation (LDA): LDA is a technique used mainly for topic modeling. You can leverage on this topic modeling to find the similarity between documents. It is assumed that more the topics two documents overlap, more are the chances that those documents carry semantic similarity.

You can study LDA in the following paper:

You can implement LDA using Gensim:

Word2Vec:

Word2Vec bring words into vector space, where words with similar semantic meaning are embedded near each other. So when plotted in a higher dimensional vector space, similar words tend to come together. The best part with Word2Vec is that it also captures semantic similarity.

You can read the original Word2Vec paper here:

You can also check the implementation in tensorflow at:

Both LDA and Word2Vec techniques can be combined to achieve interesting results. Keep following this space as we will report our findings in future blog posts.

When we look at the results achieved by such techniques, it feels that the AI is thinking. 

For a detailed understanding of Word Embeddings please refer to the following article - An Introduction to Word Embeddings


Comments

Popular posts from this blog

Implement XOR in Tensorflow

XOR is considered as the 'Hello World' of Neural Networks. It seems like the best problem to try your first TensorFlow program.

Tensorflow makes it easy to build a neural network with few tweaks. All you have to do is make a graph and you have a neural network that learns the XOR function.

Why XOR? Well, XOR is the reason why backpropogation was invented in the first place. A single layer perceptron although quite successful in learning the AND and OR functions, can't learn XOR (Table 1) as it is just a linear classifier, and XOR is a linearly inseparable pattern (Figure 1). Thus the single layer perceptron goes into a panic mode while learning XOR – it can't just do that. 

Deep Propogation algorithm comes for the rescue. It learns an XOR by adding two lines L1 and L2 (Figure 2). This post assumes you know how the backpropogation algorithm works.



Following are the steps to implement the neural network in Figure 3 for XOR in Tensorflow:
1. Import necessary libraries
impo…

Understanding Generative Adversarial Networks - Part II

In "Understanding Generative Adversarial Networks - Part I" you gained a conceptual understanding of how GAN works. In this post let us get a mathematical understanding of GANs.
The loss functions can be designed most easily using the idea of zero-sum games. 
The sum of the costs of all players is 0. This is the Minimax algorithm for GANs
Let’s break it down.
Some terminology: V(D, G) : The value function for a minimax game E(X) : Expectation of a random variable X, also equal to its average value D(x) : The discriminator output for an input x from real data, represents probability G(z): The generator's output when its given z from the noise distribution D(G(z)): Combining the above, this represents the output of the discriminator when 
given a generated image G(z) as input
Now, as explained above, the discriminator is the maximizer and hence it tries to 
maximize