Skip to main content

Science journalism meets artificial intelligence "Robotic Journalism"

By - Raghuram Vadapalli, Bakhtiyar Syed, Nishant Prabhu , Balaji Vasan Srinivasan, Vasudeva Varma.
Summary of Research papers from IIT Hyd





 Since couple of years an exciting topic is getting attraction in Machine learning and Artificial world that is "Robot Reporter".  Today's paper got inspired by the concept. Application to science journalism is non-trivial, as that would entail understanding scientific content and translating it to simpler language without distorting underlying semantics. paper heads infant steps towards answering few challenges.

 Authors came out with a tool, which, given the title and abstract of a research paper will generate a blog title by mimicking a human science journalist. The tool uses model trained on 87,328 pairs of research papers and their related blogs.



Contributions can be summed up as follows


1. A new parallel corpus of 87, 328 pairs of research paper titles and abstracts and their corresponding blog titles.

2. Demonstrating the web application, which uses a pipeline-based architecture that can generate blog titles in a step-by-step fashion,while enabling the user to choose between various heuristic functions as well as the neural model to be used for generating the blog title.

3. Analyzing the outcomes of the experiments conducted to find the best heuristic function as well as network architecture.

Architecture 





Stage 1 


 uses heuristic function to analyse and extract sequence

 What is Heuristic function...

 The Heuristic function is a way to inform the search about the direction to a goal. It provides an informed way to guess which neighbor of a node will lead to goal. There is nothing magical about heuristic function. It must use only information that can be readily obtained about node.

Stage 2 


 The pointer-generator model used to generate the output sequence from the intermediate sequences.

 Sequence to Sequence (seq2seq) is a learning model that converts an input sequence into output sequence. Seq2Seq model has achieved great success in fields such as machine translation, dialog systems, question-answering.

Blog Title Generation

Heuristic functions takes title and abstract of research paper as input H(pt, abs) where pt is paper title and abs is paper abstract. Various heuristic functions were explored and are outlined below
1)pt
2)RP (TF-IDF based)
3)RD (Flesch reading ease based)
4)RPD (normalized of RD and RP)
The output of the previous step is fed into a sequence-to-sequence neural generation model in order to generate the title of the blog post.

 System provides a baseline attention network which defines 'attention' over the input sequence to allow the network to focus on specific parts of the input text and the pointer-generator
The sequence s obtained from the first stage is the input to the neural natural language generation model which generates bt' as output with loss function  L(bt, bt'), given by sum of cross entropy loss at all time-steps:








Working prototype gives opportunity to play around with combination of heuristic functions and model types for generating blog title

 Link for working site


Labels:

Comments

Post a Comment

Popular posts from this blog

DDSP: DIFFERENTIABLE DIGITAL SIGNAL PROCESSING

 -By Google Research {jesseengel,hanoih,gcj,adarob}@google.com Paper Link Code Demo Abstract Most generative models of audio directly generate samples in one of two domains: time or frequency. While sufficient to express any signal, these representations are inefficient, as they do not utilize existing knowledge of how sound is generated and perceived. A third approach (vocoders/synthesizers) successfully incorporates strong domain knowledge of signal processing and perception, but has been less actively researched due to limited expressivity and difficulty integrating with modern auto-differentiation-based machine learning methods. In this paper, we introduce the Differentiable Digital Signal Processing (DDSP) library, which enables direct integration of classic signal processing elements with deep learning methods. Focusing on audio synthesis, we achieve high-fidelity generation without the need for large autoregressive models or adversarial losses, demonstrating that DDSP enable...
Post a Comment
Read more

Capsule Neural Networks

-By Sara Sabour, Nicholas Frosst,  Geoffrey E. Hinton  Google Brain Toronto Paper Link Abstract   A capsule is a group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part. We use the length of the activity vector to represent the probability that the entity exists and its orientation to represent the instantiation parameters. Active capsules at one level make predictions, via transformation matrices, for the instantiation parameters of higher-level capsules. When multiple predictions agree, a higher level capsule becomes active. We show that a discrimininatively trained, multi-layer capsule system achieves state-of-the-art performance on MNIST and is considerably better than a convolutional net at recognizing highly overlapping digits. To achieve these results we use an iterative routing-by-agreement mechanism: A lower-level capsule prefers to send its output to higher...
1 comment
Read more

Detecting Anomaly in Big Data System Logs Using Convolutional Neural Network

By - Siyang Lu, Xiang Wei, Yandong Li, Liqiang Wang Department of Computer Science, University of Central Florida, Orlando, FL, USA School of Software Engineering, Beijing Jiaotong University, China Paper Link Previous week we did a scan of the survey of Anomaly detection   , that covers breath of the topic (Types of Anomalies, Types of Models and their applications). In this summary paper, we look in to the core Heavy lifting part of Detecting Anomaly in Big Data System Logs Using Convolutional Neural Network. Abstract Nowadays, big data systems are being widely adopted by many domains for offering effective data solutions, such as manufacturing, healthcare, education, and media. Big data systems produce tons of unstructured logs that contain buried valuable information. However, it is a daunting task to manually unearth the information and detect system anomalies. A few automatic methods have been developed, where the cutting-edge machine lear...
Post a Comment
Read more