Author: christoffer-bjorkskog

Posted on

Tensorflow Object Detection API

Google has released an opensource framework built on top of Tensorflow, called the Tensorflow Object Detection API which is a tool for making it easy to make and deploy object detection models.

There are different state of the art types of models you can build. It you for instance make models using the Single Shot Multibox Detector (SSD) with MobileNets you will get lightweight models that you can run in real time on mobile devices.

The models you get are Single Shot Multiboc Detector, using MobileNets or Inception V2, RegionBased Fully Convolutional Networks with Resnet 101, and Faster RCMM with Resnet 101 or Inception Resnet v2.

You also get a Jupyter notebook for trying things out

 

If all these terms above makes no sense, you can read this excellent blog post explaining Deep Learning for Object Detection by Joyce Xu.

Posted on

Why fast.ai switched from Keras and Tensorflow to Pytorch and built their own Framework on top of it

In the new fast.ai course they will be using pytorch instead of Tensorflow, and has built a framework on top of it to make it even easier to use than Keras.

Pytorch is a dynamic instead of static deep learning library and Jeremy Writes that nearly all of the top 10 Kaggle competition winners now have been using Pytorch.

In the part 2 of fast.ai course the focus was to allow student so read and implement recent research papers, and pytorch made this easier due to its flexibility. It allowed them to try out things you could not do as easily with Tensorflow. It also makes it easier to understand what is going on in the algorithms as with Tensorflow, the computation becomes a black box once you send it to the GPU.

Most models trains faster on Pytorch than on Tensorflow and are easier to debug contributing to faster development iterations.

The reason they built a framework on top of Pytorch is that pytorch comes with less defaults than Keras. They want the course one to be accessible for students with little or no experience in Machine learning. Also they wanted to help avoid common pitfalls (such as not shuffling the data when needed to or vice versa) and get you going much faster, improving where Keras was lacking. They also built in many best practices that Keras was lacking. Jeremy writes that:

“We built models that are faster, more accurate, and more complex than those using Keras, yet were written with much less code.” – Jeremy Howard

The approach is to encapsulate all important data choices such as preprocessing, data augmentation, test/training/validation sets, multiclass/singleclass classification, regression and so on into Object-Oriented Classes.

“Suddenly, we were dramatically more productive, and made far less errors, because everything that could be automated, was automated.” – Jeremy Howard

Jeremy thinks that deep learning will see the same kind of library/framework explosion that front end developers have been used to during that last years. So the library you learn today will probably be obsolete in a year or two.

99,3% accuracy on dogs and cats with 3 lines of code is not bad:

Posted on

How To Learn Fast

In two days i was able to listen through half of cs231n in my spare time by listening on the youtube videos with higher than normal speed.

Nowadays i always listen to youtube videos with 2x or 3x speed.

With normal settings you can st the speed up to 2x. If you want to get the video faster than that you need to add a plugin or bookmarklet to achieve that.

You can drag these liks to your bookmarks bar, and get them as speed buttons to adjust the speed of your youtube videos…

x1 x2 x2.5 x3 x3.25 x3.5 x4

Also, check out this video on how to learn advanced concepts fast:

Posted on

Hardware for DNN Tutorial slides

The Energy-Efficient Multimedia Systems (EEMS) group at MIT has a Tutorial on Hardware Architectures for Deep Neural Networks. Here is the website: http://eyeriss.mit.edu/tutorial.html

All Slides in one PDF.

The slides explain a lot of convolutional Nerual Networks and how they work. It also describes many different topics ranging from the architectures for the winners of ImageNet and how their success also correlate with the use of GPUs for processing. The later part is more on the details in computation, what is computed where and how.

If you wish to check out each individual topic, they are also splitted into several different slides:

  • Background of Deep Neural Networks [ slides ]
  • Survey of DNN Development Resources [ slides ]
  • Survey of DNN Hardware [ slides ]
  • DNN Accelerator Architectures [ slides ]
  • Advanced Technology Opportunities [ slides ]
  • Network and Hardware Co-Design [ slides ]
  • Benchmarking Metrics [ slides ]
  • Tutorial Summary [ slides ]
  • References [ slides ]
Posted on

Want to train a deep neural net for a self driving car?

If you don’t have the time or money to spend on Udacitys Self Driving Car nanodegree, perhaps you want to try anyway to make a car drive by itself. Perhaps your real car is not ideal for training the algorithms, then you can use the provided simulator provided by Udacity that runs in Unity.

https://github.com/udacity/self-driving-car-sim

Check this blog pos tour for some tips: Training a deep learning model to steer a car in 99 lines of code

Posted on

Stanford CS231n 2017 – Convolutional Neural Networks for Visual Recognition

The video lectures for Stanfords very popular CS231n (Convolutional Neural Networks for Visual Recognition) that was held in Spring 2017 was released this month. (According to their twitter page, the cs231n website gets over 10 000 views per day. The reading material on their page is really good at explaining CNNs)

Here are the video lectures:

 

These are the assignments for the course:

 

Also Make sure to check out last years student reports.  note: one is about improving the state of the art of detecting the Higgs Boson.

Posted on

The maths you will need for AI/Machine Learning

This Hacker News thread discusses why and what kind of maths you will need if you pursue AI/Machine learning.

Here is a short summary, and i tend to agree. These where mandatory maths courses when i studied CS :

You need to have a solid foundation in:

Good  to know:

  • Graph theory or Discrete math. (no course on khan academy for that, but on great courses, which isn’t free)

Here are some books:

I like the following quote motivating why you for instance will need calculus:

Calculus essentially discusses how things change smoothly and it has a very nice mechanism for talking about smooth changes algebraically.
A system which is at an optimum will, at that exact point, be no longer increasing or decreasing: a metal sheet balanced at the peak of a hill rests flat.
Many problems in ML are optimization problems: given some set of constraints, what choices of unknown parameters minimizes error? This can be very hard (NP-hard) in general, but if you design your situation to be “smooth” then you can use calculus and its very nice set of algebraic solutions. – Commend by used Tel

It could bee very motivating for students when they first start with calculus, linear algebra and statistics if they have an idea in what fields they can practically use them later on.

Posted on

Dark Knowledge – Gueffrey Hinton

Abstract: A simple way to improve classification performance is to average the predictions of a large ensemble of different classifiers. This is great for winning competitions but requires too much computation at test time for practical applications such as speech recognition. In a widely ignored paper in 2006, Caruana and his collaborators showed that the knowledge in the ensemble could be transferred to a single, efficient model by training the single model to mimic the log probabilities of the ensemble average. This technique works because most of the knowledge in the learned ensemble is in the relative probabilities of extremely improbable wrong answers. For example, the ensemble may give a BMW a probability of one in a billion of being a garbage truck but this is still far greater (in the log domain) than its probability of being a carrot. This “dark knowledge”, which is practically invisible in the class probabilities, defines a similarity metric over the classes that makes it much easier to learn a good classifier. I will describe a new variation of this technique called “distillation” and will show some surprising examples in which good classifiers over all of the classes can be learned from data in which some of the classes are entirely absent, provided the targets come from an ensemble that has been trained on all of the classes. I will also show how this technique can be used to improve a state-of-the-art acoustic model and will discuss its application to learning large sets of specialist models without overfitting. This is joint work with Oriol Vinyals and Jeff Dean.

Video: https://www.youtube.com/watch?v=EK61htlw8hY&t=3m14s
Lecture notes: http://www.ttic.edu/dl/dark14.pdf

Posted on

Paper on Deep Reinforcement Learning

Paper: A Brief Survey of Deep Reinforcement Learning
Authors: Kai Arulkumaran, Marc Peter Deisenroth, Miles Brundage, Anil Anthony Bharath
Submitted: Submitted on 19 Aug 2017
Read the PDF

Abstract:

Deep reinforcement learning is poised to revolutionise the field of AI and represents a step towards building autonomous systems with a higher level understanding of the visual world. Currently, deep learning is enabling reinforcement learning to scale to problems that were previously intractable, such as learning to play video games directly from pixels. Deep reinforcement learning algorithms are also applied to robotics, allowing control policies for robots to be learned directly from camera inputs in the real world. In this survey, we begin with an introduction to the general field of reinforcement learning, then progress to the main streams of value-based and policy-based methods. Our survey will cover central algorithms in deep reinforcement learning, including the deep Q-network, trust region policy optimisation, and asynchronous advantage actor-critic. In parallel, we highlight the unique advantages of deep neural networks, focusing on visual understanding via reinforcement learning. To conclude, we describe several current areas of research within the field.