SpatialSim: Recognizing Spatial Configurations of Objects with Graph Neural Networks

• URL: http://arxiv.org/abs/2004.04546v2
• Date: Thu, 16 Jul 2020 18:16:31 GMT
• Title: SpatialSim: Recognizing Spatial Configurations of Objects with Graph Neural Networks
• Authors: Laetitia Teodorescu, Katja Hofmann, and Pierre-Yves Oudeyer
• Abstract summary: We show how a machine can learn and compare classes of geometric spatial configurations that are invariant to the point of view of an external observer. We propose SpatialSim (Spatial Similarity), a novel geometrical reasoning benchmark, and argue that progress on this benchmark would pave the way towards a general solution. Secondly, we study how inductive relational biases exhibited by fully-connected message-passing Graph Neural Networks (MPGNNs) are useful to solve those tasks.
• Score: 31.695447265278126
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recognizing precise geometrical configurations of groups of objects is a key capability of human spatial cognition, yet little studied in the deep learning literature so far. In particular, a fundamental problem is how a machine can learn and compare classes of geometric spatial configurations that are invariant to the point of view of an external observer. In this paper we make two key contributions. First, we propose SpatialSim (Spatial Similarity), a novel geometrical reasoning benchmark, and argue that progress on this benchmark would pave the way towards a general solution to address this challenge in the real world. This benchmark is composed of two tasks: Identification and Comparison, each one instantiated in increasing levels of difficulty. Secondly, we study how relational inductive biases exhibited by fully-connected message-passing Graph Neural Networks (MPGNNs) are useful to solve those tasks, and show their advantages over less relational baselines such as Deep Sets and unstructured models such as Multi-Layer Perceptrons. Finally, we highlight the current limits of GNNs in these tasks.

Related papers

• DepWiGNN: A Depth-wise Graph Neural Network for Multi-hop Spatial Reasoning in Text [52.699307699505646]
  We propose a novel Depth-Wise Graph Neural Network (DepWiGNN) to handle multi-hop spatial reasoning. Specifically, we design a novel node memory scheme and aggregate the information over the depth dimension instead of the breadth dimension of the graph. Experimental results on two challenging multi-hop spatial reasoning datasets show that DepWiGNN outperforms existing spatial reasoning methods.
  arXiv  Detail & Related papers  (2023-10-19T08:07:22Z)
• Memorization with neural nets: going beyond the worst case [5.662924503089369]
  In practice, deep neural networks are often able to easily interpolate their training data. For real-world data, however, one intuitively expects the presence of a benign structure so that already occurs at a smaller network size than suggested by memorization capacity. We introduce a simple randomized algorithm that, given a fixed finite dataset with two classes, with high probability constructs an interpolating three-layer neural network in time.
  arXiv  Detail & Related papers  (2023-09-30T10:06:05Z)
• A singular Riemannian geometry approach to Deep Neural Networks II. Reconstruction of 1-D equivalence classes [78.120734120667]
  We build the preimage of a point in the output manifold in the input space. We focus for simplicity on the case of neural networks maps from n-dimensional real spaces to (n - 1)-dimensional real spaces.
  arXiv  Detail & Related papers  (2021-12-17T11:47:45Z)
• Graph Neural Networks with Learnable Structural and Positional Representations [83.24058411666483]
  A major issue with arbitrary graphs is the absence of canonical positional information of nodes. We introduce Positional nodes (PE) of nodes, and inject it into the input layer, like in Transformers. We observe a performance increase for molecular datasets, from 2.87% up to 64.14% when considering learnable PE for both GNN classes.
  arXiv  Detail & Related papers  (2021-10-15T05:59:15Z)
• A neural anisotropic view of underspecification in deep learning [60.119023683371736]
  We show that the way neural networks handle the underspecification of problems is highly dependent on the data representation. Our results highlight that understanding the architectural inductive bias in deep learning is fundamental to address the fairness, robustness, and generalization of these systems.
  arXiv  Detail & Related papers  (2021-04-29T14:31:09Z)
• Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges [50.22269760171131]
  The last decade has witnessed an experimental revolution in data science and machine learning, epitomised by deep learning methods. This text is concerned with exposing pre-defined regularities through unified geometric principles. It provides a common mathematical framework to study the most successful neural network architectures, such as CNNs, RNNs, GNNs, and Transformers.
  arXiv  Detail & Related papers  (2021-04-27T21:09:51Z)
• Learning Spatial Context with Graph Neural Network for Multi-Person Pose Grouping [71.59494156155309]
  Bottom-up approaches for image-based multi-person pose estimation consist of two stages: keypoint detection and grouping. In this work, we formulate the grouping task as a graph partitioning problem, where we learn the affinity matrix with a Graph Neural Network (GNN) The learned geometry-based affinity is further fused with appearance-based affinity to achieve robust keypoint association.
  arXiv  Detail & Related papers  (2021-04-06T09:21:14Z)
• Neural Architecture Search in Graph Neural Networks [1.2881413375147996]
  This paper compares two NAS methods for optimizing Graph Neural Networks (GNN) Results consider 7 datasets over two search spaces and show that both methods obtain similar accuracies to a random search.
  arXiv  Detail & Related papers  (2020-07-31T21:04:24Z)
• The impossibility of low rank representations for triangle-rich complex networks [9.550745725703292]
  We argue that such graph embeddings do notcapture salient properties of complex networks. We mathematically prove that any embedding that can successfully create these two properties must have rank nearly linear in the number of vertices. Among other implications, this establishes that popular embedding techniques such as Singular Value Decomposition and node2vec fail to capture significant structural aspects of real-world complex networks.
  arXiv  Detail & Related papers  (2020-03-27T20:57:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.