Sunday, December 31, 2017

The European Quantum Technologies Roadmap

An updated summary of the 150-page QT Roadmap: link

Sunday, December 3, 2017

The bricklaying bot

Bricklaying robot lays 1,000+ bricks per hour! The Caterpillar-backed bot could be in production by the end of year. more

Friday, November 17, 2017

CapsuleNet

Are we at the dawn of a new era in convolutional networks?
Is it possible to combine CapsuleNet with CortexNet?

__________________________________
News:
Technology Review
ML Review


__________________________________
Notable papers:
2017
Dynamic Routing Between Capsules
Sara Sabour, Nicholas Frosst, Geoffrey Hinton
more

(Under review; submitted to ICLR 2018)
Matrix capsules with EM routing
more

2014
Optimizing Neural Networks that Generate Images
Tijmen Tieleman
See page 26

2011
Transforming Auto-encoders
Geoffrey Hinton, Alex Krizhevsky, Sida D. Wang
more

2006
To Recognize Shapes, First Learn to Generate Images
Geoffrey Hinton
See page 4, "fifth strategy"

1995
The wake-sleep algorithm for unsupervised neural networks
Geoffrey Hinton, Peter Dayan, Brendan Frey, Radford Neal
more

1981
Shape representation in parallel systems
Geoffrey Hinton
more

1979
Some demonstrations of the effects of structural descriptions in mental imagery
Geoffrey Hinton
more

1958
Oliver Selfridge
Pandemonium: A paradigm for learning
See also


__________________________________
Video:

2017
Aurélien Géron
Capsule Networks (CapsNet) - Tutorial
more

How to implement CapsNets using TensorFlow
more

2014
Geoffrey Hinton
What is wrong with Convolutional Neural Networks?
more

2012
Geoffrey Hinton: "Does the Brain do Inverse Graphics?"
more


__________________________________
Blogs:
Adrian Colyer
Dynamic routing between capsules
more

Max Pechyonkin
Understanding Hinton’s Capsule Networks. Part I: Intuition
more

Understanding Hinton’s Capsule Networks. Part II: How Capsules Work
more

Understanding Hinton’s Capsule Networks. Part III: Dynamic Routing Between Capsules
more

Jonathan Hui
Understanding Dynamic Routing between Capsules (Capsule Network)
more

Kendrick Tan
Capsule Networks Explained
more

Mike Ross
A Visual Representation of Capsule Network Computations
more

Debarko De
What is a CapsNet or Capsule Network?
more

Thibault Neveu
Understand and apply CapsNet on Traffic sign classification
more

__________________________________
Discussion:
dynamic_routing_between_capsules
capsule_networks_explained
a_visual_representation_of_capsule_network

How to treat Hinton's thesis "Dynamic Routing Between Capsules"? (Chinese)
more

__________________________________
Implementation:
CapsuleNet on TensorFlow on MNIST
CapsuleNet in Keras on MNIST
CapsuleNet in Keras on Cifar dataset
CapsuleNet in TensorFlow on German traffic sign dataset
CapsuleNet in TensorFlow on MNIST
CapsuleNet in Kaggle on MNIST

Wednesday, July 19, 2017

Amazon Alexa’s Ashwin Ram talks AI with Sebastian Thrun

Required viewing for anyone interested in the past, present, and future of AI: here

Wednesday, July 12, 2017

Public LiDAR data set

New York University releases a large laser scanned data set for urban development.

Tuesday, July 11, 2017

Motion vectors, inertia, and strategy

For a deeper understanding of vectors, inertia, and motion strategy, see Racetrack:
Try not to crash online here.
Further discussion here.

Folding Building Technology


Scalable, shape-shifting structures: here and here

Saturday, July 8, 2017

Autonomous Driving

FYI: an open-access overview of autonomous driving

Sunday, April 9, 2017

Algorithm

One of the earliest-known entries of the word "algorithm" in a dictionary can be found in Samuel Johnson's Dictionary of the English Language from 1785.



The entry on page 130 reads (approximately): algorithm: Arabic words, which are used to imply the six operations of arithmetic, or the science of numbers.

Today, "algorithm" is defined as:
- A well-defined procedure to solve a problem.
- A clearly specified set of simple instructions to be followed to solve a problem.
- A finite, deterministic, and effective problem-solving method.
- An explicit, precise, unambiguous, mechanically-executable sequence of elementary instructions.1
- A sequence of computational steps that transform input into output.2
- A set of rules that precisely defines a sequence of operations.3
- A precisely defined set of mathematical or logical operations for the performance of a particular task.4
- A step-by-step process to efficiently reach the desired goal.5
- A systematic method for solving a problem.6
- A well-defined computational procedure that takes a variable input and halts with an output.7
- A step-by-step list of instructions for solving any instance of the problem that might arise8

1. Jeff Erickson's algorithms notes
2. Introduction to Algorithms / Cormen, Leiserson, Rivest, Stein / MIT Press link
3. Introduction to Computer Organization and Data Structures (1972 ed.) / Stone / McGraw-Hill, New York.
4. "algorithm, n." OED Online, Oxford University Press, June 2019 link
5. Arizona State University graduate course syllabus "Foundations of Algorithms" CSE 551
6. Universal Book of Mathematics / Darling / Wiley / 2004
7. Handbook of Applied Cryptography / Menezes et al / CRC Press / 2001 / p.57
8. Problem Solving with AlgoDS / Miller / Univ. Auckland / 2013 / p.4 link

See also: wikipedia

Monday, January 9, 2017

Applying Deep Learning to Detect Damage in Urban Areas

Analyze and estimate the extent of damage in a disaster area: Link