‘Getting Deep into Things’: Deep Mapping in a ‘Vacant’ Landscape

Areas in cities typically denoted as ‘Vacant and Derelict Land’ are frequently presented in policy documents as absent of meaning and awaiting development. However, visits to many of these sites offer evidence of abundant citizen activity occurring outside of planning policy. Dog walkers, DIY skatepark builders, pigeon fanciers and reminiscing former factory workers, for example, can all be found inscribing their own narratives, in palimpsest like fashion, upon these landscapes. This spatio-temporally bound and layered mix of contested meanings extend beyond representational capacity offered by traditional cartographic methods as employed in policy decision-making. Such a failure to represent these ecologies of citizen-led practices often results in their erasure at the point of formal redevelopment. In this chapter, we explore how one alternative approach may respond to these challenges of representation through a case study project in Glasgow, Scotland. Deep mapping is an ethnographically informed, arts research practice, drawing Cifford Geertz’s notion of ‘thick description’ into a visual-performative realm and seeking to extend beyond the thin map by creating multifaceted and open-ended descriptions of place. As such, deep maps are not only investigations into place but of equal concern are the processes by which representations of place are generated. Implicit in this are questions about the role of the researcher as initiator, gatherer, archivist or artist and the intertwining between the place and the self. As a methodological approach that embraces multiplicity and favours the ‘politicized, passionate, and partisan’ over the totalizing objectivity of traditional maps, deep mapping offers a potential to give voice to marginalized, micro-narratives existing in tension with one another and within dominant meta-narratives but also triggers new questions over inclusivity. This methodologically focused chapter explores the ways in which an ethnographically informed, arts research practice may offer alternative insight into spaces of non-aligned narratives. The results from this investigation will offer new framings of spaces within the urban landscape conventionally represented as vacant or empty and generate perspectives on how art research methods may provide valuable investigative tools for decision-makers working in such contexts. The deep mapping work is available to view at http://www.govandeepmap.com.

Multitask learning over shared subspaces

This paper uses constructs from machine learning to define pairs of learning tasks that either shared or did not share a common subspace. Human subjects then learnt these tasks using a feedback-based approach and we hypothesised that learning would be boosted for shared subspaces. Our findings broadly supported this hypothesis with either better performance on the second task if it shared the same subspace as the first, or positive correlations over task performance for shared subspaces. These empirical findings were compared to the behaviour of a Neural Network model trained using sequential Bayesian learning and human performance was found to be consistent with a minimal capacity variant of this model. Networks with an increased representational capacity, and networks without Bayesian learning, did not show these transfer effects. We propose that the concept of shared subspaces provides a useful framework for the experimental study of human multitask and transfer learning.

Ignorance matters

The ability to reason about ignorance is an important and often overlooked representational capacity. Phillips and colleagues assume that knowledge representations are inevitably accompanied by ignorance representations. We argue that this is not necessarily the case, as agents who can reason about knowledge often fail on ignorance tasks, suggesting that ignorance should be studied as a separate representational capacity.

Long-Term Visual Robot Localization using Computational Models of the Neocortex

Visual place recognition is an essential capability for autonomous mobile robots which use cameras as their primary sensors. Although there has been a considerable amount of research in the topic, the high degree of image variability poses extra research challenges. Following advances in neuroscience, new biologically inspired models have been developed. Inspired by the human neocortex, hierarchical temporal memory model has potential to identify temporal sequences of spatial patterns using sparse distributed representations, which are known to have high representational capacity and high tolerance to noise. These features are interesting for place recognition applications. Some authors have proposed simplifications from the original framework, such as starting from an empty set of minicolumns and increasing the number of minicolumns on demand instead of the usage of a fixed number of minicolumns whose connections adapt over time. In this paper, we investigate the usage of framework originally proposed with the aim of extending the run-time during long-term operations. Results show that the proposed architecture can encode an internal representation of the world using a fixed number of cells in order to improve system scalability.

Reifying Representations

The representational theory of mind (RTM) holds that the mind is stocked with mental representations: mental items that represent. They can be stored in memory, manipulated during mental activity, and combined to form complex representations. RTM is widely presupposed within cognitive science, which offers many successful theories that cite mental representations. Nevertheless, mental representations are still viewed warily in some scientific and philosophical circles. This chapter develops a novel version of RTM: the capacities-based representational theory of mind (C-RTM). According to C-RTM, a mental representation is an abstract type that marks the exercise of a representational capacity. Talk about mental representations embodies an ontologically loaded way of classifying mental states through representational capacities that the states deploy. Complex mental representations mark the appropriate joint exercise of multiple representational capacities. The chapter supports C-RTM with examples drawn from cognitive science, including perceptual representations and cognitive maps, and applies C-RTM to long-standing debates over the existence, nature, individuation, structure, and explanatory role of mental representations.

EVALUATION OF UNet AND UNet++ ARCHITECTURES IN HIGH RESOLUTION IMAGE CHANGE DETECTION APPLICATIONS

Change detection applications from satellite imagery can be a very useful tool in monitoring human activities and understanding their interaction with the physical environment. In the past few years most of the recent research approaches to automatic change detection have been based on the application of Deep Learning techniques and especially on variations of Convolutional Neural Network architectures due to their great representational capacity and their state-of-the-art performance in visual tasks such as image classification and semantic segmentation. In this work we train and evaluate two CNN architectures, UNet and UNet++, on a change detection task using Very High-Resolution satellite images collected at two different time epochs. We also examine and analyse the effect of two different loss functions, a combination of the Binary Cross Entropy Loss with the Dice Loss, and the Lovász Hinge loss, both of which were specifically designed for semantic segmentation applications. Finally, we experiment with the use of data augmentation as well as deep supervision techniques to evaluate and quantify their contribution in the final classification performance of the different network architectures.

Participatory Representation in a Non-Western Context: The Case of Homeowner Associations in Beijing

This article examines the concept and practices of “participatory representation” in the Chinese context, a subset of substantive representation that emphasizes “authenticity” and participatory engagement in solving neighborhood problems. Through examining Chinese homeowner associations (HOAs), we explain how representation operates at the neighborhood level in a grassroots organizational context without a Western style of democracy, identifying the determinants and capacities of participatory representation. By proposing a model of representational capacity and using logistic regression analysis, we find that four factors have an impact on the quality of participatory representation: (1) homeowner attributes (i.e., gender, occupation, and length of residence), (2) problem-solving effectiveness of representative organizations, (3) transparent and open elections, and (4) level of homeowner participation. We further suggest that in a transitional society like China, these representative organizations, namely, HOAs, act as important training grounds for democratic skills, through which participatory citizen engagement is being learned and cultivated. This study contributes to contemporary accounts of participatory representation by identifying the informal representation patterns within HOAs and their potential to foster civic participation and social democracy in China in the coming decades.

The Comparison of Animals

This chapter discusses the historical context of Kant’s theory of animal minds. The continuity thesis is discussed. This is the claim that, whatever the variations in their mental lives, animal and human minds manifest no differences in kind but rather exhibit the same general type of mental capacities merely exercised with very different degrees of sophistication. Kant is an ardent denier of the continuity thesis in that he claims that human beings are different in kind from animals by virtue of our ability for self-conscious understanding and the opportunities for normative self-determination that this ability affords. The approaches of Montaigne, Descartes, and Bayle are outlined. It is claimed that the relevant cognitive achievement with which Kant was concerned was that of the comparison of representations with each other and the noting of similarity or difference. It is argued that Kant adopted an analogy strategy, which claims that animals possess a capacity for the comparison of representations that is only analogous to human beings’ representational capacity.

Learning Graph Convolutional Network for Skeleton-Based Human Action Recognition by Neural Searching

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.

Abstraction and Representational Capacity in Computational Structures

Michael Weisberg’s book Simulation and Similarity argued that although mathematical models are sometimes described in narrative form, they are best understood as interpreted mathematical structures. But how can a mathematical structure be causal, as many models described in narrative seem to be? This chapter argues that models with apparently narrative form are actually computational structures. It explores this suggestion in detail, examining what computational structure consists of, the resources it offers modelers, and why attempting to re-describe computational models as imaginary concrete systems fails even more dramatically than it does for mathematical models.