I am a PhD student in the Department of Mechanical Engineering at IIT Kharagpur, India. Presently, I am working on machinery fault diagnosis and prognosis using deep learning. During my PhD, I developed an interest in machine learning and deep learning in particular. Though my training is in mechanical engineering, I have acquired machine learning skills by self-study and from MOOCs through online certifications. Beyond research, I like literature and music.
PhD in Mechanical Engineering, (Continuing)
Indian Institute of Technology, Kharagpur, India
MTech in Mechanical Engineering, 2015
National Institute of Technology, Rourkela, India
BTech in Mechanical Engineering, 2013
College of Engineering and Technology, Bhubaneswar, India
Run in Google Colab View source on GitHub Download notebook In this post, we will discuss about IndexedSlices class of Tensorflow. We will try to answer the following questions in this blog: What are IndexedSlices? Where do we get it? How to convert from IndexedSlices to tensors? What are IndexedSlices? According to Tensorflow documentation, IndexedSlices are sparse representation of a set of tensor slices at a given index.
View GitHub Page ----- View source on GitHub Download code (.zip) This code has been merged with D2L book. See PR: 1756, 1768 This post contains Tensorflow 2 code for Attention Mechanisms chapter of Dive into Deep Learning (D2L) book. The chapter has 7 sections and code for each section can be found at the following links.
Run in Google Colab View source on GitHub Download notebook In this post, we will read multiple csv files using Tensroflow Sequence. In an earlier post we had demonstrated the procedure for reading multiple csv files using a custom generator. Though generators are convenient for handling chunks of data from a large dataset, they have limited portability and scalability (see the caution here).
Run in Google Colab View source on GitHub Download notebook In many engineering applications data are usually stored in CSV (Comma Separated Values) files. In big data applications, it’s not uncommon to obtain thousands of csv files. As the number of files increases, at some point, we can no longer load the whole dataset into computer’s memory. In deep learning applications it is increasingly common to come across datasets that don’t fit in the computer’s memory.
Note: Whether this method is efficient or not is contestable. Efficiency of a data input pipeline depends on many factors. How efficiently data are loaded? What is the computer architecture on which computations are being done? Is GPU available? And the list goes on. So readers might get different performance results when they use this method in their own problems. For the simple (and small) problem considered in this post, we got no perceivable performance improvement.