High-level Inference Research Articles

Fully integrated modelling for sustainability assessment of resource recovery from waste

This paper presents an integrated modelling approach for value assessments, focusing on resource recovery from waste. The method tracks and forecasts a range of values across environmental, social, economic and technical domains by attaching these to material-flows, thus building upon and integrating unidimensional models such as material flow analysis (MFA) and lifecycle assessment (LCA). We argue that the usual classification of metrics into these separate domains is useful for interpreting the outputs of multidimensional assessments, but unnecessary for modelling. We thus suggest that multidimensional assessments can be better performed by integrating the calculation methods of unidimensional models rather than their outputs. To achieve this, we propose a new metric typology that forms the foundation of a multidimensional model. This enables dynamic simulations to be performed with material-flows (or values in any domain) driven by changes in value in other domains. We then apply the model in an illustrative case highlighting links between the UK coal-based electricity-production and concrete/cement industries, investigating potential impacts that may follow the increased use of low-carbon fuels (biomass and solid recovered fuels; SRF) in the former. We explore synergies and trade-offs in value across domains and regions, e.g. how changes in carbon emissions in one part of the system may affect mortality elsewhere. This highlights the advantages of recognising complex system dynamics and making high-level inferences of their effects, even when rigorous analysis is not possible. We also indicate how changes in social, environmental and economic ‘values’ can be understood as being driven by changes in the technical value of resources. Our work thus emphasises the advantages of building fully integrated models to inform conventional sustainability assessments, rather than applying hybrid approaches that integrate outputs from parallel models. The approach we present demonstrates that this is feasible and lays the foundations for such an integrated model.

Towards Bayesian Deep Learning: A Framework and Some Existing Methods

While perception tasks such as visual object recognition and text understanding play an important role in human intelligence, subsequent tasks that involve inference, reasoning, and planning require an even higher level of intelligence. The past few years have seen major advances in many perception tasks using deep learning models. For higher-level inference, however, probabilistic graphical models with their Bayesian nature are still more powerful and flexible. To achieve integrated intelligence that involves both perception and inference, it is naturally desirable to tightly integrate deep learning and Bayesian models within a principled probabilistic framework, which we call Bayesian deep learning. In this unified framework, the perception of text or images using deep learning can boost the performance of higher-level inference and in return, the feedback from the inference process is able to enhance the perception of text or images. This paper proposes a general framework for Bayesian deep learning and reviews its recent applications on recommender systems, topic models, and control. In this paper, we also discuss the relationship and differences between Bayesian deep learning and other related topics such as the Bayesian treatment of neural networks.

Learning to Make Inferences in a Semantic Parsing Task

We introduce a new approach to training a semantic parser that uses textual entailment judgements as supervision. These judgements are based on high-level inferences about whether the meaning of one sentence follows from another. When applied to an existing semantic parsing task, they prove to be a useful tool for revealing semantic distinctions and background knowledge not captured in the target representations. This information is used to improve the quality of the semantic representations being learned and to acquire generic knowledge for reasoning. Experiments are done on the benchmark Sportscaster corpus (Chen and Mooney, 2008), and a novel RTE-inspired inference dataset is introduced. On this new dataset our method strongly outperforms several strong baselines. Separately, we obtain state-of-the-art results on the original Sportscaster semantic parsing task.

Efficient Cache Management Model for Cooperative Reasoning in Wide Area Surveillance

For intelligent wide area surveillance, distributed cooperative inference between agents should be provided efficiently. In this paper, efficient cache management model for better context inference based on distributed data infrastructure is investigated. Context aware computing with inference based on ontology is widely used in distributed surveillance environment. However even smart devices, they generally have small memory and power which can manage only part of surveillance data. Higher level inference considering related wide area information can cause huge data transmission. Surveillance data integration merges and aligns ontologies according to the semantic similarity for efficient cooperation. For such a collaborative network, ICN(Information Centric Network) is adopted and an effective cache management scheme that can help the efficient ontology integration. In this paper, we propose an efficient cache management model which is adaptive to the actual device demands. Our scheme shows the efficiency of model resulted in better and efficient context inference.

Hierarchy-associated semantic-rule inference framework for classifying indoor scenes

Typically, the initial task of classifying indoor scenes is challenging, because the spatial layout and decoration of a scene can vary considerably. Recent efforts at classifying object relationships commonly depend on the results of scene annotation and predefined rules, making classification inflexible. Furthermore, annotation results are easily affected by external factors. Inspired by human cognition, a scene-classification framework was proposed using the empirically based annotation (EBA) and a match-over rule-based (MRB) inference system. The semantic hierarchy of images is exploited by EBA to construct rules empirically for MRB classification. The problem of scene classification is divided into low-level annotation and high-level inference from a macro perspective. Low-level annotation involves detecting the semantic hierarchy and annotating the scene with a deformable-parts model and a bag-of-visual-words model. In high-level inference, hierarchical rules are extracted to train the decision tree for classification. The categories of testing samples are generated from the parts to the whole. Compared with traditional classification strategies, the proposed semantic hierarchy and corresponding rules reduce the effect of a variable background and improve the classification performance. The proposed framework was evaluated on a popular indoor scene dataset, and the experimental results demonstrate its effectiveness.

Fusing data in adaptive agent control systems for electrical grids

This paper reports a Monte Carlo analysis of a novel multi-agent technique for network-centric control in a microgrid scenario. Agents monitor a feeder line shared by n microgrids and collectively assess changes in the direction and magnitude of power system dynamic stability. Every agent samples voltage variation every few milliseconds, classifies observations, and votes to express and communicate their individual inferences about the state of system stability. Monte Carlo simulations are used to investigate the probability that rules for representing and fusing information reliably produce error-resilient collective outcomes (ERCOs) in centralised and decentralised networks. ERCOs overcome communications or decision-making errors to provide a window of opportunity for adapting to correct emergent instability or to minimise harm. Voting systems provide a semantics and syntax for relating low-level data to high-level inferences about network situations. Our results provide a basis for optimising the design of communications infrastructure to control electrical grid stability.

What do we mean by prediction in language comprehension?

ABSTRACTWe consider several key aspects of prediction in language comprehension: its computational nature, the representational level(s) at which we predict, whether we use higher-level representations to predictively pre-activate lower level representations, and whether we “commit” in any way to our predictions, beyond pre-activation. We argue that the bulk of behavioural and neural evidence suggests that we predict probabilistically and at multiple levels and grains of representation. We also argue that we can, in principle, use higher-level inferences to predictively pre-activate information at multiple lower representational levels. We suggest that the degree and level of predictive pre-activation might be a function of its expected utility, which, in turn, may depend on comprehenders’ goals and their estimates of the relative reliability of their prior knowledge and the bottom-up input. Finally, we argue that all these properties of language understanding can be naturally explained and productively explored within a multi-representational hierarchical actively generative architecture whose goal is to infer the message intended by the producer, and in which predictions play a crucial role in explaining the bottom-up input.

Human behaviour recognition in data-scarce domains

This paper presents the novel theory for performing multi-agent activity recognition without requiring large training corpora. The reduced need for data means that robust probabilistic recognition can be performed within domains where annotated datasets are traditionally unavailable. Complex human activities are composed from sequences of underlying primitive activities. We do not assume that the exact temporal ordering of primitives is necessary, so can represent complex activity using an unordered bag. Our three-tier architecture comprises low-level video tracking, event analysis and high-level inference. High-level inference is performed using a new, cascading extension of the Rao–Blackwellised Particle Filter. Simulated annealing is used to identify pairs of agents involved in multi-agent activity. We validate our framework using the benchmarked PETS 2006 video surveillance dataset and our own sequences, and achieve a mean recognition F-Score of 0.82. Our approach achieves a mean improvement of 17% over a Hidden Markov Model baseline.

Detecting disease variants in case-parent trio studies using the bioconductor software package trio.

Case-parent trio studies are commonly employed in genetics to detect variants underlying common complex disease risk. Both commercial and freely available software suites for genetic data analysis usually contain methods for case-parent trio designs. A user might, however, experience limitations with these packages, which can include missing functionality to extend the software if a desired analysis has not been implemented, and the inability to programmatically capture all the software versions used for low-level processing and high-level inference of genomic data, a critical consideration in particular for high-throughput experiments. Here, we present a software vignette (i.e., a manual with step by step instructions and examples to demonstrate software functionality) for reproducible genome-wide analyses of case-parent trio data using the open source Bioconductor package trio. The workflow for the practitioner uses data from previous genetic trio studies to illustrate functions for marginal association tests, assessment of parent-of-origin effects, power and sample size calculations, and functions to detect gene-gene and gene-environment interactions associated with disease.

A FIRST STAGE COMPARATIVE SURVEY ON HUMAN ACTIVITY RECOGNITION METHODOLOGIES

The development of vision-based human activity recognition and analysis systems has been a matter of great interest to both the research community and practitioners during the last 20 years. Traditional methods that require a human operator watching raw video streams are nowadays deemed as at least ineffective and expensive. New, smart solutions in automatic surveillance and monitoring have emerged, propelled by significant technological advances in the fields of image processing, artificial intelligence, electronics and optics, embedded computing and networking, molding the future of several applications that can benefit from them, like security and healthcare. The main motivation behind it is to exploit the highly informative visual data captured by cameras and perform high-level inference in an automatic, ubiquitous and unobtrusive manner, so as to aid human operators, or even replace them. This survey attempts to comprehensively review the current research and development on vision-based human activity recognition. Synopses from various methodologies are presented in an effort to garner the advantages and shortcomings of the most recent state-of-the-art technologies. Also a first-level self-evaluation of methodologies is also proposed, which incorporates a set of significant features that best describe the most important aspects of each methodology in terms of operation, performance and others and weighted by their importance. The purpose of this study is to serve as a reference for further research and evaluation to raise thoughts and discussions for future improvements of each methodology towards maturity and usefulness.

Adapting Social Neuroscience Measures for Schizophrenia Clinical Trials, Part 2: Trolling the Depths of Psychometric Properties

The psychometric properties of 4 paradigms adapted from the social neuroscience literature were evaluated to determine their suitability for use in clinical trials of schizophrenia. This 2-site study (University of California, Los Angeles and University of North Carolina) included 173 clinically stable schizophrenia outpatients and 88 healthy controls. The social cognition battery was administered twice to the schizophrenia group (baseline, 4-week retest) and once to the control group. The 4 paradigms included 2 that assess perception of nonverbal social and action cues (basic biological motion and emotion in biological motion) and 2 that involve higher level inferences about self and others' mental states (self-referential memory and empathic accuracy). Each paradigm was evaluated on (1) patient vs healthy control group differences, (2) test-retest reliability, (3) utility as a repeated measure, and (4) tolerability. Of the 4 paradigms, empathic accuracy demonstrated the strongest characteristics, including large between-group differences, adequate test-retest reliability (.72), negligible practice effects, and good tolerability ratings. The other paradigms showed weaker psychometric characteristics in their current forms. These findings highlight challenges in adapting social neuroscience paradigms for use in clinical trials.

Adapting social neuroscience measures for schizophrenia clinical trials, part 3: fathoming external validity.

It is unknown whether measures adapted from social neuroscience linked to specific neural systems will demonstrate relationships to external variables. Four paradigms adapted from social neuroscience were administered to 173 clinically stable outpatients with schizophrenia to determine their relationships to functionally meaningful variables and to investigate their incremental validity beyond standard measures of social and nonsocial cognition. The 4 paradigms included 2 that assess perception of nonverbal social and action cues (basic biological motion and emotion in biological motion) and 2 that involve higher level inferences about self and others' mental states (self-referential memory and empathic accuracy). Overall, social neuroscience paradigms showed significant relationships to functional capacity but weak relationships to community functioning; the paradigms also showed weak correlations to clinical symptoms. Evidence for incremental validity beyond standard measures of social and nonsocial cognition was mixed with additional predictive power shown for functional capacity but not community functioning. Of the newly adapted paradigms, the empathic accuracy task had the broadest external validity. These results underscore the difficulty of translating developments from neuroscience into clinically useful tasks with functional significance.

Bayesian characterization of uncertainty in intra-subject non-rigid registration

In settings where high-level inferences are made based on registered image data, the registration uncertainty can contain important information. In this article, we propose a Bayesian non-rigid registration framework where conventional dissimilarity and regularization energies can be included in the likelihood and the prior distribution on deformations respectively through the use of Boltzmann's distribution. The posterior distribution is characterized using Markov Chain Monte Carlo (MCMC) methods with the effect of the Boltzmann temperature hyper-parameters marginalized under broad uninformative hyper-prior distributions. The MCMC chain permits estimation of the most likely deformation as well as the associated uncertainty. On synthetic examples, we demonstrate the ability of the method to identify the maximum a posteriori estimate and the associated posterior uncertainty, and demonstrate that the posterior distribution can be non-Gaussian. Additionally, results from registering clinical data acquired during neurosurgery for resection of brain tumor are provided; we compare the method to single transformation results from a deterministic optimizer and introduce methods that summarize the high-dimensional uncertainty. At the site of resection, the registration uncertainty increases and the marginal distribution on deformations is shown to be multi-modal.

LEARNING A COMPOSITIONAL REPRESENTATION FOR FACADE OBJECT CATEGORIZATION

Abstract. Our objective is the categorization of the most dominant objects in facade images, like windows, entrances and balconies. In order to execute an image interpretation of complex scenes we need an interaction between low level bottom-up feature detection and highlevel inference from top-down. A top-down approach would use results of a bottom-up detection step as evidence for some high-level inference of scene interpretation. We present a statistically founded object categorization procedure that is suited for bottom-up object detection. Instead of choosing a bag of features in advance and learning models based on these features, it is more natural to learn which features best describe the target object classes. Therefore we learn increasingly complex aggregates of line junctions in image sections from man-made scenes. We present a method for the classification of image sections by using the histogram of diverse types of line aggregates.

Speaker Diarization: A Review of Recent Research

Speaker diarization is the task of determining “who spoke when?” in an audio or video recording that contains an unknown amount of speech and also an unknown number of speakers. Initially, it was proposed as a research topic related to automatic speech recognition, where speaker diarization serves as an upstream processing step. Over recent years, however, speaker diarization has become an important key technology for many tasks, such as navigation, retrieval, or higher level inference on audio data. Accordingly, many important improvements in accuracy and robustness have been reported in journals and conferences in the area. The application domains, from broadcast news, to lectures and meetings, vary greatly and pose different problems, such as having access to multiple microphones and multimodal information or overlapping speech. The most recent review of existing technology dates back to 2006 and focuses on the broadcast news domain. In this paper, we review the current state-of-the-art, focusing on research developed since 2006 that relates predominantly to speaker diarization for conference meetings. Finally, we present an analysis of speaker diarization performance as reported through the NIST Rich Transcription evaluations on meeting data and identify important areas for future research.

A Taxonomy of Agent Technologies for Ubiquitous Computing Environments

The design, development and deployment of Mobile Agent (MA) systems for high-level inference and surveillance in wireless sensor networks and RFID systems have drawn increasing attention in the past decade. To answer how the state-of-the-art of MA in a wide range of ubiquitous and sensor environments is, this paper investigates the current progress of MA. It proposes a taxonomy, by which MA systems in ubiquitous computing environments are decomposed and discussed. Then, this paper provides insights into the strengths and weaknesses of existing efforts. Finally, it presents a series of solutions from the viewpoint of various roles of MA in ubiquitous environments and situations.

Design and Development of MSA agent for Rectangular and U Slotted MSA

research interest has increased in the design, development, and deployment of microstrip antenna agent systems for high-level inference and surveillance in a wireless sensor network (WSN). The proposed Antenna agent systems employ migrating codes to facilitate flexible application re- tasking, local processing, and calculating parameters. This provides extra flexibility, as well as new capabilities to MSA. It employs the genetic Algorithm to generate various results. This provides extra flexibility as well as new capability to design Rectangular and U slotted Microstrip Antenna.The development of Rectangular & U slit loaded antenna into Algorithm design based on parameter analysis and the coding for the architecture of rectangular and U slot loaded MSA agent. The taxonomy covers low level to high level design issue and facilitate the variation in design parameters like h, fr, Ɛ r, flow, fhigh, R according to the requirement of the proposed agent. IndexTerms-- Microstrip antenna, communication, dielectric constant, frequency, bandwidth, java program, algorithm,Optimization program, genetic algorithm, SHF, U slot.

Knowledge representation and inference techniques to improve the management of gas and oil facilities

This paper describes an experimental work carried out in the framework of an important European project to create and make use of a wide-ranging knowledge base in the gas/oil domain. In the context of this work, “knowledge base” means a collection of formal statement relating, with a negligible loss of information, the inner content (the ‘meaning’) of “complex events” included in two different “storyboards”. These events – originally presented under the form of unstructured natural language information – concern some general activities proper to the management of gas/oil facilities, like recognizing and monitoring gas leakage alarms in a gas processing plant or triggering the different steps needed to activate a gas turbine. To express this sort of information and to set up the knowledge base, the NKRL (Narrative Knowledge Representation Language) formalism has been used. NKRL is a conceptual meta-model and Computer Science environment expressly created to deal, in an ‘intelligent’ and complete way, with complex and content-rich ‘narrative’ data sources. The final knowledge base has been firstly tested in depth using the standard NKRL querying and information retrieval tools. High-level inference procedures have then been used, both “transformation rules” – unsuccessful queries are ‘transformed’ to produce results that are ‘semantically similar’ to those searched for initially – and “hypothesis rules” – information in the knowledge base is automatically aggregated to supply a sort of ‘causal’ explanation of some retrieved events.

The intermediate effect in clinical case recall is present in musculoskeletal physiotherapy

Previous research into medical expertise has shown that expert physicians perform significantly better than non-experts when providing a diagnosis for a clinical case. By contrast, when asked to recall the details of the case, individuals of intermediate level of expertise (typically, final-year students) obtain better results than novices and experienced physicians, a phenomenon called the “intermediate effect”. However, this effect has not been documented outside of internal medicine and dentistry. The aim of this paper was to investigate the possibility that the intermediate effect in clinical case recall exists in musculoskeletal physiotherapy. 40 participants (sport science students, musculoskeletal physiotherapy students and expert musculoskeletal physiotherapists) were assigned to four groups based upon their experience in musculoskeletal physiotherapy (control group with no experience, novices, intermediates and experts). Participants were instructed to read a case study, provide a diagnosis, justify this diagnosis, and recall the case study. It was found that the quality of diagnosis and the number of high-level inferences increased with expertise. By contrast, recall performance followed an inverted U-curve, with the best recall being obtained by intermediates. Thus, the intermediate effect is present in musculoskeletal physiotherapy, despite clear differences in the way medicine and musculoskeletal physiotherapy are taught.

Intent Inference and Syntactic Tracking with GMTI Measurements

In conventional target tracking systems, human operators use the estimated target tracks to make higher level inference of the target behaviour/intent. The work presented here develops syntactic filtering algorithms that assist human operators by extracting spatial patterns from target tracks to identify suspicious/anomalous spatial trajectories. The targets' spatial trajectories are modeled by a stochastic context free grammar (SCFG) and a switched mode state space model. Bayesian filtering algorithms for SCFGs are presented for extracting the syntactic structure and illustrated for a ground moving target indicator (GMTI) radar example. The performance of the algorithms is tested with the experimental data collected using DRDC Ottawa's X-band Wideband Experimental Airborne Radar (XWEAR).