skip to main content

Title: Learning Action-Effect Dynamics for Hypothetical Vision-Language Reasoning Task
Actions’ play a vital role in how humans interact with the world. Thus, autonomous agents that would assist us in everyday tasks also require the capability to perform ‘Reasoning about Actions & Change’ (RAC). This has been an important research direction in Artificial Intelligence (AI) in general, but the study of RAC with visual and linguistic inputs is relatively recent. The CLEVR_HYP is one such testbed for hypothetical vision-language reasoning with actions as the key focus. In this work, we propose a novel learning strategy that can improve reasoning about the effects of actions. We implement an encoder-decoder architecture to learn the representation of actions as vectors. We combine the aforementioned encoder-decoder architecture with existing modality parsers and a scene graph question answering model to evaluate our proposed system on the CLEVR_HYP dataset. We conduct thorough experiments to demonstrate the effectiveness of our proposed approach and discuss its advantages over previous baselines in terms of performance, data efficiency, and generalization capability.  more » « less
Award ID(s):
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
Findings of EMNLP 2022.
Page Range / eLocation ID:
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. In multi-agent domains (MADs), an agent's action may not just change the world and the agent's knowledge and beliefs about the world, but also may change other agents' knowledge and beliefs about the world and their knowledge and beliefs about other agents' knowledge and beliefs about the world. The goals of an agent in a multi-agent world may involve manipulating the knowledge and beliefs of other agents' and again, not just their knowledge/belief about the world, but also their knowledge about other agents' knowledge about the world. Our goal is to present an action language (mA+) that has the necessary features to address the above aspects in representing and RAC in MADs. mA+ allows the representation of and reasoning about different types of actions that an agent can perform in a domain where many other agents might be present -- such as world-altering actions, sensing actions, and announcement/communication actions. It also allows the specification of agents' dynamic awareness of action occurrences which has future implications on what agents' know about the world and other agents' knowledge about the world. mA+ considers three different types of awareness: full-, partial- awareness, and complete oblivion of an action occurrence and its effects. This keeps the language simple, yet powerful enough to address a large variety of knowledge manipulation scenarios in MADs. The semantics of mA+ relies on the notion of state, which is described by a pointed Kripke model and is used to encode the agent's knowledge and the real state of the world. It is defined by a transition function that maps pairs of actions and states into sets of states. We illustrate properties of the action theories, including properties that guarantee finiteness of the set of initial states and their practical implementability. Finally, we relate mA+ to other related formalisms that contribute to RAC in MADs. 
    more » « less
  2. As one of the popular deep learning methods, deep convolutional neural networks (DCNNs) have been widely adopted in segmentation tasks and have received positive feedback. However, in segmentation tasks, DCNN-based frameworks are known for their incompetence in dealing with global relations within imaging features. Although several techniques have been proposed to enhance the global reasoning of DCNN, these models are either not able to gain satisfying performances compared with traditional fully-convolutional structures or not capable of utilizing the basic advantages of CNN-based networks (namely the ability of local reasoning). In this study, compared with current attempts to combine FCNs and global reasoning methods, we fully extracted the ability of self-attention by designing a novel attention mechanism for 3D computation and proposed a new segmentation framework (named 3DTU) for three-dimensional medical image segmentation tasks. This new framework processes images in an end-to-end manner and executes 3D computation on both the encoder side (which contains a 3D transformer) and the decoder side (which is based on a 3D DCNN). We tested our framework on two independent datasets that consist of 3D MRI and CT images. Experimental results clearly demonstrate that our method outperforms several state-of-the-art segmentation methods in various metrics. 
    more » « less
  3. Abstract

    This paper introduces a new graph neural network architecture for learning solutions of Capacitated Vehicle Routing Problems (CVRP) as policies over graphs. CVRP serves as an important benchmark for a wide range of combinatorial planning problems, which can be adapted to manufacturing, robotics and fleet planning applications. Here, the specific aim is to demonstrate the significant real-time executability and (beyond training) scalability advantages of the new graph learning approach over existing solution methods. While partly drawing motivation from recent graph learning methods that learn to solve CO problems such as multi-Traveling Salesman Problem (mTSP) and VRP, the proposed neural architecture presents a novel encoder-decoder architecture. Here the encoder is based on Capsule networks, which enables better representation of local and global information with permutation invariant node embeddings; and the decoder is based on the Multi-head attention (MHA) mechanism allowing sequential decisions. This architecture is trained using a policy gradient Reinforcement Learning process. The performance of our approach is favorably compared with state-of-the-art learning and non-learning methods for a benchmark suite of Capacitated-VRP (CVRP) problems. A further study on the CVRP with demand uncertainties is conducted to explore how this Capsule-Attention Mechanism architecture can be extended to handle real-world uncertainties by embedding them through the encoder.

    more » « less
  4. Reed-Solomon (RS) codes are adopted in many digital communication and storage systems to ensure data reliability. For many of these systems, the encoder and decoder are not active at the same time. In previous designs, RS encoders implemented as linear feedback shift registers in a concatenated structure are reused to compute the syndromes so that the decoder complexity is reduced. However, the parallel versions of such encoders have very long critical path and hence can not achieve high speed. This paper proposes a new parallel resource-shareable RS encoder architecture based on the Chinese Remainder Theorem (CRT). The generator polynomial of RS codes is decomposed into factors of degree one and state transformation is developed to enable the sharing of the hardware units for syndrome computation. As a result, the critical path is reduced to only one multiplier and one adder, regardless of the parallelism. Additionally, by utilizing the property that the degrees of the decomposed polynomial factors are one, optimizations are also developed to greatly simplify the CRT-based encoder. For example encoders of a (255, 229) RS code over GF(2^8), our proposed design can achieve at least 29% higher efficiency in terms of area-time product for moderate or higher parallelisms compared to the previous resource-shareable RS encoder and traditional parallel RS encoders combined with syndrome computation units that implement the same functionality. 
    more » « less
  5. The key idea of current deep learning methods for dense prediction is to apply a model on a regular patch centered on each pixel to make pixel-wise predictions. These methods are limited in the sense that the patches are determined by network architecture instead of learned from data. In this work, we propose the dense transformer networks, which can learn the shapes and sizes of patches from data. The dense transformer networks employ an encoder-decoder architecture, and a pair of dense transformer modules are inserted into each of the encoder and decoder paths. The novelty of this work is that we provide technical solutions for learning the shapes and sizes of patches from data and efficiently restoring the spatial correspondence required for dense prediction. The proposed dense transformer modules are differentiable, thus the entire network can be trained. We apply the proposed networks on biological image segmentation tasks and show superior performance is achieved in comparison to baseline methods.

    more » « less