Abstract The dominant object detection approaches treat the recognition of each region separately and overlook crucial semantic correlations between objects in one scene. This paradigm leads to substantial performance drop when facing heavy long-tail problems, where very few samples are available for rare classes and plenty of confusing categories exists. We exploit diverse human commonsense knowledge for reasoning over large-scale object categories and reaching semantic coherency within one image. Particularly, we present Hybrid Knowledge Routed Modules (HKRM) that incorporates the reasoning routed by two kinds of knowledge forms: an explicit knowledge module for structured constraints that are summarized with linguistic knowledge (e.g. shared attributes, relationships) about concepts; and an implicit knowledge module that depicts some implicit constraints (e.g. common spatial layouts). By functioning over a region-to-region graph, both modules can be individualized and adapted to coordinate with visual patterns in each image, guided by specific knowledge forms. HKRM are light-weight, general-purpose and extensible by easily incorporating multiple knowledge to endow any detection networks the ability of global semantic reasoning. Experiments on large-scale object detection benchmarks show HKRM obtains around 34.5% improvement on VisualGenome (1000 categories) and 30.4% on ADE in terms of mAP.
A framework of new unified neural and neuro-fuzzy approaches for integrating implicit and explicit knowledge in neuro-symbolic systems is proposed. In the developed hybrid system, training data set is used for building neurofuzzy modules, and represents implicit domain knowledge. On the other hand, the explicit domain knowledge is represented by fuzzy rules, which are directly mapped into equivalent neural structures. Three methods to combine the explicit and implicit knowledge modules are proposed.
The main aim of this paper is to raise awareness of higherorder knowledge (knowledge about someone else's knowledge) as an issue for computer game AI. We argue that a number of existing game genres, especially those involving social interaction, are natural fields of application for an approach we call explicit knowledge programming. We motivate the use of this approach, and describe a simple implementation based upon it. A survey of recent literature and computer games illustrates its novelty.
Forward-thinking businesses constantly look for ways to improve their operations and grow their capabilities, which can help them to stand apart from their rivals, attract more customers, and grow your annual profit margin. It doesn't matter if you run a large or small company, you must look for ways to improve your USP, business communications, and marketing power. Get started by reading about the four learning opportunities your business should consider. While you may have an in-depth knowledge of your brand, goods or services, you might not have a firm knowledge on what it takes to manage and finance a business. If you are struggling with the operational side of running a business, but want to improve your productivity, profitability, and efficiency, consider a business administration course.
Lemoisson, Philippe (Centre de coopération internationale en recherche agronomique pour le développement (CIRAD)) | Surroca, Guillaume (Le Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier (LIRMM)) | Jonquet, Clément (Le Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier (LIRMM)) | Cerri, Stefano Alessandro (Le Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier (LIRMM))
Reconciling the ecosystem of semantic Web data with the ecosystem of social Web participation has been a major issue for the Web Science community. To answer this need, we propose an innovative approach called ViewpointS where the knowledge is topologically, rather than logically, explored and assessed. Both social contributions and linked data are represented by triples agent-resource-resource called “viewpoints”. A “viewpoint” is the subjective declaration by an agent (human or artificial) of some semantic proximity between two resources. Knowledge resources and viewpoints form a bipartite graph called “knowledge graph”. Information retrieval is processed on demand by choosing a user’s “perspective” i.e., rules for quantifying and aggregating “viewpoints” which yield a “knowledge map”. This map is equipped with a topology: the more viewpoints between two given resources, the shorter the distance; moreover, the distances between resources evolve along time according to new viewpoints, in the metaphor of synapses’ strengths. Our hypothesis is that these dynamics actualize an adaptive, actionable collective knowledge.