Learnable Higher-Order Representation for Action Recognition

Author(s):  
Jie Shao ◽  
Xiangyang Xue
Author(s):  
J. Christopher Maloney

Carruthers proposes a subtle dispositionalist rendition of higher order theory regarding phenomenal character. The theory would distinguish unconscious movement management from conscious attitude management as perceptual processes. Each process takes perceptual representations as inputs. A representation subject to attitude management is apt to induce a higher order representation of itself that secures a self-referential aspect of its content supposedly determinative of phenomenal character. Unfortunately, the account requires a problematic cognitive ambiguity while failing to explain why attitude, but not movement, management, determines character. Moreover, normal variation in attitudinal management conflicts with the constancy typical of phenomenal character. And although an agent denied perceptual access to a scene about which she is otherwise well informed would suffer no phenomenal character, dispositionalist theory entails otherwise. Such problems, together with the results of the previous chapters, suggest that, whether cloaked under intentionalism or higher order theory, representationalism mistakes content for character.


2020 ◽  
Vol 34 (03) ◽  
pp. 2669-2676 ◽  
Author(s):  
Wei Peng ◽  
Xiaopeng Hong ◽  
Haoyu Chen ◽  
Guoying Zhao

Human action recognition from skeleton data, fuelled by the Graph Convolutional Network (GCN) with its powerful capability of modeling non-Euclidean data, has attracted lots of attention. However, many existing GCNs provide a pre-defined graph structure and share it through the entire network, which can loss implicit joint correlations especially for the higher-level features. Besides, the mainstream spectral GCN is approximated by one-order hop such that higher-order connections are not well involved. All of these require huge efforts to design a better GCN architecture. To address these problems, we turn to Neural Architecture Search (NAS) and propose the first automatically designed GCN for this task. Specifically, we explore the spatial-temporal correlations between nodes and build a search space with multiple dynamic graph modules. Besides, we introduce multiple-hop modules and expect to break the limitation of representational capacity caused by one-order approximation. Moreover, a corresponding sampling- and memory-efficient evolution strategy is proposed to search in this space. The resulted architecture proves the effectiveness of the higher-order approximation and the layer-wise dynamic graph modules. To evaluate the performance of the searched model, we conduct extensive experiments on two very large scale skeleton-based action recognition datasets. The results show that our model gets the state-of-the-art results in term of given metrics.


Author(s):  
Nilam Nur Amir Sjarif ◽  
Siti Zaiton Mohd Hashim ◽  
Siti Mariyam Shamsuddin ◽  
Anca L. Ralescu

2007 ◽  
Vol 14 (14) ◽  
Author(s):  
Olivier Danvy ◽  
Michael Spivey

Over forty years ago, David Barron and Christopher Strachey published a startlingly elegant program for the Cartesian product of a list of lists, expressing it with a three nested occurrences of the function we now call <em>foldr</em>. This program is remarkable for its time because of its masterful display of higher-order functions and lexical scope, and we put it forward as possibly the first ever functional pearl. We first characterize it as the result of a sequence of program transformations, and then apply similar transformations to a program for the classical power set example. We also show that using a higher-order representation of lists allows a definition of the Cartesian product function where <em>foldr</em> is nested only twice.


Sign in / Sign up

Export Citation Format

Share Document