Snapshot Ensemble is a method to obtain multiple neural network which can be ensembled at no additional training cost. This is achieved by letting a single neural network converge into several local minima along its optimization path and save the model parameters at certain epochs, therefore the weights being ‘snapshots’ of the model.
Modern machine learning platforms like Tensorflow have to date been used mainly by the computer science crowd, for applications like computer vision and language understanding. But as JJ Allaire pointed out in his keynote at the RStudio conference earlier this month (embedded below), there’s a wealth of applications in the data science domain that have yet to be widely explored using these techniques. This includes things like time series forecasting, logistic regression, latent variable models, and censored data analysis (including survival analysis and failure data analysis).
Back then I made a proof of concept package called pysty (Python style) to make R behave a bit more like Python for importing packages and functions. I had forgotten about this package…Fast forward to this week.
Circa 1997, the reigning world chess champion Garry Kasparov was against an unknown opponent. The opponent was formidable. Garry was not playing a human. He was playing the game with IBM’s behemoth supercomputer, Deep Blue.
Text analysis, as a whole, is an emerging field of study. Fields such as Marketing, Product Management, Academia, and Governance are already leveraging the process of analyzing and extracting information from textual data. We discussed the technology behind Text Classification, one of the essential parts of Text Analysis. Text classification or Text Categorization is the activity of labeling natural language texts with relevant categories from a predefined set. In laymen terms, text classification is a process of extracting generic tags from unstructured text. These generic tags come from a set of pre-defined categories. Classifying your content and products into categories help users to easily search and navigate within website or application.
One of the reasons why I love R is that I feel like I’m constantly finding out about cool new packages through an ever-growing community of users and teachers. To understand the current state of R packages on CRAN, I ran some code provided by Gergely Daróczi on Github . As of today there have been almost 14,000 R packages published on CRAN and the rate of publishing appears to be growing at an almost exponential trend. Additionally, there are even more packages available on sources like Github, Bioconductor, Bitbucket and more.
Every once in a while, a python library is developed that has the potential of changing the landscape in the field of deep learning. PyTorch is one such library. In the last few weeks, I have been dabbling a bit in PyTorch. I have been blown away by how easy it is to grasp. Among the various deep learning libraries I have used till date – PyTorch has been the most flexible and effortless of them all. In this article, we will explore PyTorch with a more hands-on approach, covering the basics along with a case study. We will also compare a neural network built from scratch in both numpy and PyTorch to see their similarities in implementation.