In this paper we proposed PWOC-3D, a novel end-to-end CNN pipeline to predict scene flow (optical flow and stereo disparity jointly) directly from stereo image sequences. Our approach was significantly more efficient than earlier classical approac...
We introduced neural vector conceptualization as a means of interpreting continuous samples from a word vector space. We demonstrated that our method produces considerably less entropic concept activation profiles than the cosine similarity measur...
We successfully transferred a new explanation method to the NLP domain. We were able to demonstrate that PatternAttribution can be used to identify meaningful signal contributions in text inputs. Our method should be extended to other popular mode...
Credentials & highlights
- Project Views