The problem of heterogeneous case representation poses a major obstacle to realising real-life multi-case-base reasoning (MCBR) systems. The knowledge overhead in developing and maintaining ...translation protocols between distributed case bases poses a serious challenge to CBR developers. In this paper, we situate CBR as a flexible problem-solving strategy that relies on several heterogeneous knowledge containers. We introduce a technique called language games to solve the interoperability issue. Our technique has two phases. The first is an eager learning phase where case bases communicate to build a shared indexing lexicon of similar cases in the distributed network. The second is the problem-solving phase where, using the distributed index, a case base can quickly consult external case bases if the local solution is insufficient. We provide a detailed description of our approach and demonstrate its effectiveness using an evaluation on a real data set from the tourism domain.
Research in multi-objective reinforcement learning (MORL) has introduced the utility-based paradigm, which makes use of both environmental rewards and a function that defines the utility derived by ...the user from those rewards. In this paper we extend this paradigm to the context of single-objective reinforcement learning (RL), and outline multiple potential benefits including the ability to perform multi-policy learning across tasks relating to uncertain objectives, risk-aware RL, discounting, and safe RL. We also examine the algorithmic implications of adopting a utility-based approach.
Many researchers have made use of the Wikipedia network for relatedness and similarity tasks. However, most approaches use only the most recent information and not historical changes in the network. ...We provide an analysis of entity relatedness using temporal graph-based approaches over different versions of the Wikipedia article link network and DBpedia, which is an open-source knowledge base extracted from Wikipedia. We consider creating the Wikipedia article link network as both a union and intersection of edges over multiple time points and present a novel variation of the Jaccard index to weight edges based on their transience. We evaluate our results against the KORE dataset, which was created in 2010, and show that using the 2010 Wikipedia article link network produces the strongest result, suggesting that semantic similarity is time sensitive. We then show that integrating multiple time frames in our methods can give a better overall similarity demonstrating that temporal evolution can have an important effect on entity relatedness.
Typically, case-based recommender systems recommend single items to the on-line customer. In this paper we introduce the idea of recommending a user-defined collection of items where the user has ...implicitly encoded the relationships between the items. Automated collaborative filtering (ACF), a socalled ‘contentless’ technique, has been widely used as a recommendation strategy for music items. However, its reliance on a global model of the user’s interests makes it unsuited to catering for the user’s local interests. We consider the context-sensitive task of building a compilation, a user-defined collection of music tracks. In our analysis, a collection is a case that captures a specific shortterm information/music need. In an offline evaluation, we demonstrate how a case-completion strategy that uses short-term representations is significantly more effective than the ACF technique. We then consider the problem of recommending a compilation according to the user’s most recent listening preferences. Using a novel on-line evaluation where two algorithms compete for the user’s attention, we demonstrate how a knowledge-light case-based reasoning strategy successfully addresses this problem.
Pandemics or high impact epidemics are one of the biggest threats facing humanity today. While a complete elimination of the occurrence of such threats is improbable, it is possible to contain their ...impact by efficient management which in turn depends on effective decision-making. In the event of a pandemic the data flows are enormous and pose severe cognitive overload to the public health decision-makers. In this context, this paper presents PandemCap, an innovative decision support tool that can be used by the public health officials for making better and well informed decisions in the event of pandemics or high impact epidemics. PandemCap provides an interactive, flexible platform to public health decision-makers by making extensive use of techniques from the domains of visual analytics and epidemic modeling. In addition, the tool also allows for the study of the impact of various interventions or control measures such as the use of vaccines, anti-virals, hospital beds, and ventilators.
The realization that scholarly publications are discussed and have influence on discourse outside scientific and academic domains has given rise to area of scientometrics called alternative metrics ...or "altmetrics". Furthermore, researchers in this field tend to focus primarily on measuring scientific activity on social media platforms such as Twitter, however these count-based metrics are vulnerable to gaming because they tend to lack concrete justification or reference to the primary source. In this collaboration with Elsevier, we extend the conventional citation graph to a heterogeneous graph of publications, scientists, venues, organizations and more authoritative media sources such as mainstream news and weblogs. Our approach consists of two parts: one is integrating the bibliometric data with the social data such as blogs, mainstream news. The other involves understanding how standard graph-based metrics can be used to predict the academic impact. Our result showed the computed graph-based metrics can reasonably predict the academic impact of early stage researchers.
In many risk-aware and multi-objective reinforcement learning settings, the utility of the user is derived from the single execution of a policy. In these settings, making decisions based on the ...average future returns is not suitable. For example, in a medical setting a patient may only have one opportunity to treat their illness. When making a decision, just the expected return -- known in reinforcement learning as the value -- cannot account for the potential range of adverse or positive outcomes a decision may have. Our key insight is that we should use the distribution over expected future returns differently to represent the critical information that the agent requires at decision time.
In this paper, we propose Distributional Monte Carlo Tree Search, an algorithm that learns a posterior distribution over the utility of the different possible returns attainable from individual policy executions, resulting in good policies for risk-aware settings. Moreover, our algorithm outperforms the state-of-the-art in multi-objective reinforcement learning for the expected utility of the returns.
A Diffusion-Based Method for Entity Search Torres-Tramon, Pablo; Timilsina, Mohan; Hayes, Conor
2019 IEEE 13th International Conference on Semantic Computing (ICSC),
2019-Jan.
Conference Proceeding
Entity search has become an import task in the Web of Data recently. Most solutions developed so far have focused on modelling entity search using standard information retrieval model and adapting ...graph-based objects to multi-fielded pseudo-documents. Among the models proposed to this regard, we can found bag-of-words, multi-gram, and mixtures of language models. While these works have produced interesting findings, little attention has been put on the graph structure of the Web of data. In this work, we aim to fill this gap by introducing a two-stage method based on a standard information retrieval model combined with a diffusion-based approach. We implemented and tested several diffusion models finding that heat kernel diffusion processes have a competitive performance with state-of-the-art models.
Scalar Reward is Not Enough Vamplew, Peter; Smith, Benjamin J.; Källström, Johan ...
Proceedings of the 22nd International Conference on Autonomous Agents and Multiagent Systems (AAMAS),
2023
Conference Proceeding
Silver et al.14 posit that scalar reward maximisation is sufficient to underpin all intelligence and provides a suitable basis for artificial general intelligence (AGI). This extended abstract ...summarises the counter-argument from our JAAMAS paper 19.
PDF
