In Biomechanics

The Connectionists

There is a creed of deep learning where its adherants follow the principle that deep learning can be most successful by modeling the way the human brain works.


A visualization of a deep learning auto-encoder
An important part of deep learning is the process of improvement and...  well, learning which occurs.   Machine learning can occur with both unlabeled and labeled input data.  After a lot of it, the machine can learn what to learn, and actually learn.


A visualization of the deep learning process of uncovering hidden features
The Deepin deep learning comes from the idea that the process is highly abstracted.  Layers and patterns are hidden underneath each other.


Advanced machine learning techniques like those studied and developed here at biglab have important ramifications for the biomechanical simulations.  As Collectionists understand that the intricacies of the human brain should be projected to deep learning, deep learning can then be projected to biomechanical simulations to help predict things like muscle activations for mechanical tasks.



Picture of Najeeb Khan

Najeeb Khan
PhD Student