Faster-LTN: a neuro-symbolic, end-to-end object detection architecture

Abstract

The detection of semantic relationships between objects represented in an image is one of the fundamental challenges in image interpretation. Neural-Symbolic techniques, such as Logic Tensor Networks (LTNs), allow the combination of semantic knowledge representation and reasoning with the ability to efficiently learn from examples typical of neural networks. We here propose Faster-LTN, an object detector composed of a convolutional backbone and an LTN. To the best of our knowledge, this is the first attempt to combine both frameworks in an end-to-end training setting. This architecture is trained by optimizing a grounded theory which combines labelled examples with prior knowledge, in the form of logical axioms. Experimental comparisons show competitive performance with respect to the traditional Faster R-CNN architecture.

Logic Tensor Networks

“…In [3], LTN is used to annotate a reinforcement learning environment with prior knowledge and incorporate latent information into an agent. In [42], authors embed LTN in a state-of-the-art convolutional object detector. Extensions and generalizations of LTN have also been proposed in the past years, such as LYRICS [47] and Differentiable Fuzzy Logic (DFL) [68,69].…”