scrapbook
  • "Unorganized" Notes
  • The Best Public Datasets for Machine Learning and Data Science
  • Practice Coding
  • plaid-API project
  • Biotech
    • Machine Learning vs. Deep Learning
  • Machine Learning for Computer Graphics
  • Books (on GitHub)
  • Ideas/Thoughts
  • Ziva for feature animation: Stylized simulation and machine learning-ready workflows
  • Tools
  • 🪶math
    • Papers
    • Math for ML (coursera)
      • Linear Algebra
        • Wk1
        • Wk2
        • Wk3
        • Wk4
        • Wk5
      • Multivariate Calculus
    • Improving your Algorithms & Data Structure Skills
    • Algorithms
    • Algorithms (MIT)
      • Lecture 1: Algorithmic Thinking, Peak Finding
    • Algorithms (khan academy)
      • Binary Search
      • Asymptotic notation
      • Sorting
      • Insertion sort
      • Recursion
      • Solve Hanoi recursively
      • Merge Sort
      • Representing graphs
      • The breadth-first search algorithm
      • Breadth First Search in JavaScript
      • Breadth-first vs Depth-first Tree Traversal in Javascript
    • Algorithms (udacity)
      • Social Network
    • Udacity
      • Linear Algebra Refresher /w Python
    • math-notes
      • functions
      • differential calculus
      • derivative
      • extras
      • Exponentials & logarithms
      • Trigonometry
    • Probability (MIT)
      • Unit 1
        • Probability Models and Axioms
        • Mathematical background: Sets; sequences, limits, and series; (un)countable sets.
    • Statistics and probability (khan academy)
      • Analyzing categorical data
      • Describing and comparing distributions
      • Outliers Definition
      • Mean Absolute Deviation (MAD)
      • Modeling data distribution
      • Exploring bivariate numerical data
      • Study Design
      • Probability
      • Counting, permutations, and combinations
      • Binomial variables
        • Binomial Distribution
        • Binomial mean and standard deviation formulas
        • Geometric random variable
      • Central Limit Theorem
      • Significance Tests (hypothesis testing)
    • Statistics (hackerrank)
      • Mean, Medium, Mode
      • Weighted Mean
      • Quartiles
      • Standard Deviation
      • Basic Probability
      • Conditional Probability
      • Permutations & Combinations
      • Binomial Distribution
      • Negative Binomial
      • Poisson Distribution
      • Normal Distribution
      • Central Limit Theorem
      • Important Concepts in Bayesian Statistics
  • 📽️PRODUCT
    • Product Strategy
    • Product Design
    • Product Development
    • Product Launch
  • 👨‍💻coding
    • of any interest
    • Maya API
      • Python API
    • Python
      • Understanding Class Inheritance in Python 3
      • 100+ Python challenging programming exercises
      • coding
      • Iterables vs. Iterators vs. Generators
      • Generator Expression
      • Stacks (LIFO) / Queues (FIFO)
      • What does -1 mean in numpy reshape?
      • Fold Left and Right in Python
      • Flatten a nested list of lists
      • Flatten a nested dictionary
      • Traverse A Tree
      • How to Implement Breadth-First Search
      • Breadth First Search
        • Level Order Tree Traversal
        • Breadth First Search or BFS for a Graph
        • BFS for Disconnected Graph
      • Trees and Tree Algorithms
      • Graph and its representations
      • Graph Data Structure Interview Questions
      • Graphs in Python
      • GitHub Repo's
    • Python in CG Production
    • GLSL/HLSL Shading programming
    • Deep Learning Specialization
      • Neural Networks and Deep Learning
      • Untitled
      • Untitled
      • Untitled
    • TensorFlow for AI, ML, and DL
      • Google ML Crash Course
      • TensorFlow C++ API
      • TensorFlow - coursera
      • Notes
      • An Introduction to different Types of Convolutions in Deep Learning
      • One by One [ 1 x 1 ] Convolution - counter-intuitively useful
      • SqueezeNet
      • Deep Compression
      • An Overview of ResNet and its Variants
      • Introducing capsule networks
      • What is a CapsNet or Capsule Network?
      • Xception
      • TensorFlow Eager
    • GitHub
      • Project README
    • Agile - User Stories
    • The Open-Source Data Science Masters
    • Coding Challenge Websites
    • Coding Interview
      • leetcode python
      • Data Structures
        • Arrays
        • Linked List
        • Hash Tables
        • Trees: Basic
        • Heaps, Stacks, Queues
        • Graphs
          • Shortest Path
      • Sorting & Searching
        • Depth-First Search & Breadth-First Search
        • Backtracking
        • Sorting
      • Dynamic Programming
        • Dynamic Programming: Basic
        • Dynamic Programming: Advanced
    • spaCy
    • Pandas
    • Python Packages
    • Julia
      • jupyter
    • macos
    • CPP
      • Debugging
      • Overview of memory management problems
      • What are lvalues and rvalues?
      • The Rule of Five
      • Concurrency
      • Avoiding Data Races
      • Mutex
      • The Monitor Object Pattern
      • Lambdas
      • Maya C++ API Programming Tips
      • How can I read and parse CSV files in C++?
      • Cpp NumPy
    • Advanced Machine Learning
      • Wk 1
      • Untitled
      • Untitled
      • Untitled
      • Untitled
  • data science
    • Resources
    • Tensorflow C++
    • Computerphile
      • Big Data
    • Google ML Crash Course
    • Kaggle
      • Data Versioning
      • The Basics of Rest APIs
      • How to Make an API
      • How to deploying your API
    • Jupiter Notebook Tips & Tricks
      • Jupyter
    • Image Datasets Notes
    • DS Cheatsheets
      • Websites & Blogs
      • Q&A
      • Strata
      • Data Visualisation
      • Matplotlib etc
      • Keras
      • Spark
      • Probability
      • Machine Learning
        • Fast Computation of AUC-ROC score
    • Data Visualisation
    • fast.ai
      • deep learning
      • How to work with Jupyter Notebook on a remote machine (Linux)
      • Up and Running With Fast.ai and Docker
      • AWS
    • Data Scientist
    • ML for Beginners (Video)
    • ML Mastery
      • Machine Learning Algorithms
      • Deep Learning With Python
    • Linear algebra cheat sheet for deep learning
    • DL_ML_Resources
    • Awesome Machine Learning
    • web scraping
    • SQL Style Guide
    • SQL - Tips & Tricks
  • 💡Ideas & Thoughts
    • Outdoors
    • Blog
      • markdown
      • How to survive your first day as an On-set VFX Supervisor
    • Book Recommendations by Demi Lee
  • career
    • Skills
    • learn.co
      • SQL
      • Distribution
      • Hypothesis Testing Glossary
      • Hypothesis Tests
      • Hypothesis & AB Testing
      • Combinatorics Continued and Maximum Likelihood Estimation
      • Bayesian Classification
      • Resampling and Monte Carlo Simulation
      • Extensions To Linear Models
      • Time Series
      • Distance Metrics
      • Graph Theory
      • Logistic Regression
      • MLE (Maximum Likelihood Estimation)
      • Gradient Descent
      • Decision Trees
      • Ensemble Methods
      • Spark
      • Machine Learning
      • Deep Learning
        • Backpropagation - math notation
        • PRACTICE DATASETS
        • Big Data
      • Deep Learning Resources
      • DL Datasets
      • DL Tutorials
      • Keras
      • Word2Vec
        • Word2Vec Tutorial Part 1 - The Skip-Gram Model
        • Word2Vec Tutorial Part 2 - Negative Sampling
        • An Intuitive Explanation of Convolutional Neural Networks
      • Mod 4 Project
        • Presentation
      • Mod 5 Project
      • Capstone Project Notes
        • Streaming large training and test files into Tensorflow's DNNClassifier
    • Carrier Prep
      • The Job Search
        • Building a Strong Job Search Foundation
        • Key Traits of Successful Job Seekers
        • Your Job Search Mindset
        • Confidence
        • Job Search Action Plan
        • CSC Weekly Activity
        • Managing Your Job Search
      • Your Online Presence
        • GitHub
      • Building Your Resume
        • Writing Your Resume Summary
        • Technical Experience
      • Effective Networking
        • 30 Second Elevator Pitch
        • Leveraging Your Network
        • Building an Online Network
        • Linkedin For Research And Networking
        • Building An In-Person Network
        • Opening The Line Of Communication
      • Applying to Jobs
        • Applying To Jobs Online
        • Cover Letters
      • Interviewing
        • Networking Coffees vs Formal Interviews
        • The Coffee Meeting/ Informational Interview
        • Communicating With Recruiters And HR Professional
        • Research Before an Interview
        • Preparing Questions for Interviews
        • Phone And Video/Virtual Interviews
        • Cultural/HR Interview Questions
        • The Salary Question
        • Talking About Apps/Projects You Built
        • Sending Thank You's After an Interview
      • Technical Interviewing
        • Technical Interviewing Formats
        • Code Challenge Best Practices
        • Technical Interviewing Resources
      • Communication
        • Following Up
        • When You Haven't Heard From an Employer
      • Job Offers
        • Approaching Salary Negotiations
      • Staying Current in the Tech Industry
      • Module 6 Post Work
      • Interview Prep
  • projects
    • Text Classification
    • TERRA-REF
    • saildrone
  • Computer Graphics
  • AI/ML
  • 3deeplearning
    • Fast and Deep Deformation Approximations
    • Compress and Denoise MoCap with Autoencoders
    • ‘Fast and Deep Deformation Approximations’ Implementation
    • Running a NeuralNet live in Maya in a Python DG Node
    • Implement a Substance like Normal Map Generator with a Convolutional Network
    • Deploying Neural Nets to the Maya C++ API
  • Tools/Plugins
  • AR/VR
  • Game Engine
  • Rigging
    • Deformer Ideas
    • Research
    • brave rabbit
    • Useful Rigging Links
  • Maya
    • Optimizing Node Graph for Parallel Evaluation
  • Houdini
    • Stuff
    • Popular Built-in VEX Attributes (Global Variables)
Powered by GitBook
On this page
  1. coding
  2. TensorFlow for AI, ML, and DL

Deep Compression

PreviousSqueezeNetNextAn Overview of ResNet and its Variants

Last updated 6 years ago

is by Song Han, Huizi Mao, and William J. Dally. It won the best paper award at ICLR 2016. It details three methods of compressing a neural network in order to reduce the size of the network on disk, improve performance, and decrease run time.

Pre-trained convolutional neural networks are too large for mobile devices: is 240 MB and is over 552 MB. This seems small when compared to a music library or large video, but the difference is that the networks reside in memory when running. On mobile devices SRAM is scarce and DRAM is expensive to access in terms of energy used. For reference, the authors estimate that a 1 billion node network running at 20 FPS on your phone would draw nearly 13 Watts from just the DRAM access alone. If these networks are going to run on a mobile device (where they could, for example, automatically tag pictures as they are taken) they must be compressed in some manner. In this paper the authors apply three compression methods to the weights of various networks and measure the results. A diagram of the three methods and their results are below; I’ll walk you through them in more depth in the next few paragraphs.A summary of the three stages in the compression pipeline proposed by Han, Mao, and Dally. Note that the size reduction is cumulative. Image from their paper.

The first compression method is Network Pruning. In this method a network is fully trained and then any connections with a weight below a certain threshold are removed leaving a sparse network. The sparse network is then retrained to ensure the remaining connections are used optimally. This form of compression reduced the size of AlexNet by a factor of 9, and VGG-16 by a factor of 13. The authors also use a clever data structure that makes use of variably sized integers to store the network after this compression.

The second compression method is Trained Quantization and Weight Sharing. Here the weights in a network are clustered together with other weights of similar magnitude, and all these weights are then represented by a single shared value. The authors use k-means clustering to group weights for sharing. They explore multiple methods of setting the k centroids and find that a simple linear spacing of the centroids along the full distribution of weight values performs best. This compression method reduces the size of the networks by a factor of 3 or 4. A diagram with an example of this compression technique is shown below.A toy example of trained quantization and weight sharing. On the top row, weights of the same color have been clustered and will be replaced by a centroid value. On the bottom row, gradients are calculated and used to update the centroids. From Han, Mao, and Dally’s paper.

Using all three compression methods leads to a compression factor of 35 times for AlexNet, and 49 times for VGG-16! This reduces AlexNet to 6.9 MB, and VGG-16 to under 11.3 MB! Unsurprisingly it is the fully connected layers that are the largest (90% of the model size), but they also compress the best (96% of weights pruned in VGG-16). The new, smaller convolutional layers run faster than their old versions (4 times faster on mobile GPU) and use less energy (4 times less). These results are achieved with no loss in performance! A plot showing the energy efficiency and speedups due to compression are shown below:The energy efficiency and speedups due to compression for various layers in the neural networks. The dense bars are the results before compression, and the pruned bars are the results after. Note the Y axis is log10! From Han, Mao, and Dally’s paper.

Han, Mao, and Dally’s compression techniques achieve an almost perfect result: the in memory size of a network is reduced, the run speed is increased, and the energy used to perform the calculation is decreased. Although designed with mobile in mind, their compression is so successful I would not be surprised to see it widely supported by the various deep learning frameworks in the near future.

The third and final compression method is Huffman Coding. Huffman coding is a standard lossless compression technique. The general idea is that it uses fewer bits to represent data that appears frequently and more bits to represent data that appears infrequently. For more details see the . Huffman coding reduces network size by 20% to 30%.

👨‍💻
Wikipedia Article
The next paper from our reading group
AlexNet
VGG-16