How transferable are features in deep neural networks?
- J. Yosinski, J. Clune, Yoshua Bengio, Hod Lipson
- Computer ScienceNIPS
- 6 November 2014
This paper quantifies the generality versus specificity of neurons in each layer of a deep convolutional neural network and reports a few surprising results, including that initializing a network with transferred features from almost any number of layers can produce a boost to generalization that lingers even after fine-tuning to the target dataset.
Distilling Free-Form Natural Laws from Experimental Data
- Michael D. Schmidt, Hod Lipson
- PhysicsScience
- 3 April 2009
This work proposes a principle for the identification of nontriviality, and demonstrated this approach by automatically searching motion-tracking data captured from various physical systems, ranging from simple harmonic oscillators to chaotic double-pendula, and discovered Hamiltonians, Lagrangians, and other laws of geometric and momentum conservation.
Understanding Neural Networks Through Deep Visualization
- J. Yosinski, J. Clune, Anh M Nguyen, Thomas J. Fuchs, Hod Lipson
- Computer ScienceArXiv
- 22 June 2015
This work introduces several new regularization methods that combine to produce qualitatively clearer, more interpretable visualizations of convolutional neural networks.
Fabricated: The New World of 3D Printing
- Hod Lipson, M. Kurman
- Art
- 11 February 2013
Fabricated tells the story of 3D printers, humble manufacturing machines that are bursting out of the factory and into homes, businesses, schools, kitchens, hospitals, even the fashion catwalk.The…
Resilient Machines Through Continuous Self-Modeling
- J. Bongard, V. Zykov, Hod Lipson
- BiologyScience
- 17 November 2006
A robot is described that can recover from change autonomously, through continuous self-modeling, and this concept may help develop more robust machines and shed light on self- modeling in animals.
Universal robotic gripper based on the jamming of granular material
- E. Brown, Nicholas Rodenberg, H. Jaeger
- Computer ScienceProceedings of the National Academy of Sciences
- 22 September 2010
It is shown that volume changes of less than 0.5% suffice to grip objects reliably and hold them with forces exceeding many times their weight, and opens up new possibilities for the design of simple, yet highly adaptive systems that excel at fast gripping of complex objects.
Modular Self-Reconfigurable Robot Systems [Grand Challenges of Robotics]
- Mark H. Yim, Wei-Min Shen, G. Chirikjian
- Computer ScienceIEEE robotics & automation magazine
- 2 April 2007
Several of the key directions for the future of modular self-reconfigurable robotic systems, including the design, fabrication, motion planning, and control of autonomous kinematic machines with variable morphology are shown.
The evolutionary origins of modularity
- J. Clune, Jean-Baptiste Mouret, Hod Lipson
- BiologyProceedings of the Royal Society B: Biological…
- 11 July 2012
It is demonstrated that the ubiquitous, direct selection pressure to reduce the cost of connections between network nodes causes the emergence of modular networks.
Automatic design and manufacture of robotic lifeforms
- Hod Lipson, J. Pollack
- EngineeringNature
- 31 August 2000
A combined computational and experimental approach is reported in which simple electromechanical systems are evolved through simulations from basic building blocks, and the ‘fittest’ machines are then fabricated robotically using rapid manufacturing technology.
Automated reverse engineering of nonlinear dynamical systems
- J. Bongard, Hod Lipson
- Computer ScienceProceedings of the National Academy of Sciences
- 12 June 2007
This work introduces for the first time a method that can automatically generate symbolic equations for a nonlinear coupled dynamical system directly from time series data, applicable to any system that can be described using sets of ordinary nonlinear differential equations.
...
...