Super interesting paper from Fei-fei Li group tackles scene graph generation

[Yuhuang Hu]

The paper is here: https://arxiv.org/pdf/1701.02426v1.pdf

Super interesting results, one can take an image and then generate a full scene graph for inference.

A great step to fix sub-symbolic-symbolic gap.

Yuhuang.

[Ben Goertzel]

Interesting. This approach shares a lot of ideas, at the high level,
with the approach we're pursuing, that builds a scene graph from the
patterns mined from the state of a DeSTIN-like deep NN.... The
contextual biasing of object recognition is an important part of our
approach and they are doing that too, in their own way...

ben

> --
> You received this message because you are subscribed to the Google Groups
> "opencog" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to opencog+u...@googlegroups.com.
> To post to this group, send email to ope...@googlegroups.com.
> Visit this group at https://groups.google.com/group/opencog.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/opencog/3c7646b3-613c-4e80-9fed-df09eeb80984%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



--
Ben Goertzel, PhD
http://goertzel.org

“I tell my students, when you go to these meetings, see what direction
everyone is headed, so you can go in the opposite direction. Don’t
polish the brass on the bandwagon.” – V. S. Ramachandran

[Yuhuang Hu]

Emmm... I see... It's quite interesting in recent developments of unsupervised learning also.

GAN seems to be a promising approach in feature generation. Have you considered to try?

Yuhuang.

[Ben Goertzel]

We have been looking at stacked InfoGAN, which is sort of close to our
goals with DeSTIN... I think a pyramidal infoGAN might do something
like what I want...

> https://groups.google.com/d/msgid/opencog/34964f6e-9518-4832-bc13-912e20c99828%40googlegroups.com.

[Yuhuang Hu]

Interesting... Could be helpful if one can integrate Convolutional GRU in order to exploit temporal relations between layers  (https://arxiv.org/pdf/1511.06432v4.pdf)