Inference In Probabilistic Models Research Articles

Monitoring of work execution processes in railway construction using expert technologies

The article presents the abstract theorem and practical options for monitoring the technological process of a railway track object construction using expert technologies. Railway construction as a complex dynamic system requires certain resources for its maintenance. Under these conditions, effective control over the work execution technology performed during the construction of railway track facilities is of decisive importance. This can be achieved by improving the existing engineering and technical support system for railway construction through the introduction of a subsystem for engineering and technological processes intellectual support for the railway facilities construction. One of the subsystem tasks is the effective use of automation tools with elements of artificial intelligence. The appearance of deviations from the planned requirements during the construction work execution, due to the stochastic nature of railway construction, leads to technology violations, a decrease in productivity, an increase in labor costs, prime cost, an increase in duration and, as a result, a deterioration in the railway track object operational characteristics. In order to avoid such a development of events, periodic monitoring of the technological solutions implementation is necessary. To this end, as part of the methodology formation for engineering and intellectual support of railway construction technological processes, a system for monitoring the technological processes implementation for the railway facilities construction has been developed. Monitoring includes three stages: parameters registration, assessment and forecast of the technological process implementation. Taking into account the formalization peculiarities of a number of tasks to be solved, a method for registering the indicators of the technological process implementation on a time interval was developed and described. The results obtained at the assessment stage made it possible to determine the possibilities of using an expert system built on a probabilistic inference model for processing the results of monitoring the work progress during the railway facility construction. The logical inference mechanism is organized on a system of feedback and inference based on probabilistic Bayesian logic. The forecast of the technological process implementation is based on general methodological issues of the Markov random processes application. The predictive model and the method for obtaining quantitative characteristics of the forecast are based on the interpretation of the simplest flow properties in order to analyze the dynamics of changes in its states. Based on the results of a theoretical study, the article presents the practical aspects of monitoring the work execution processes in railway construction using expert technologies on the example of the construction of a flooded railway subgrade roadbed. The results presented in the article were obtained in the course of a dissertation research carried out by the author.

Intellectualization of monitoring and regulation of the technological processes implementation in railway construction

The article presents abstract theorems and practical options for monitoring and regulating the technological process implementation of building a railway track object using artificial intelligence methods and tools. Railway construction as a complex dynamic system requires certain resources for its maintenance. Under these conditions, effective control over the construction work technology is of decisive importance. This can be achieved by improving the existing engineering and technical support system for railway construction through the introduction of an engineering and intellectual support subsystem for technological processes for railway facilities construction. One of the tasks of the subsystem is the effective use of automation tools with artificial intelligence elements. The deviations occurrence from planned requirements during construction work, due to the stochastic nature of railway construction, leads to violations of technology, an increase in labor costs, prime cost, an increase in duration, and as a result, a deterioration in the operational railway facility characteristics. To avoid such a development of events, a prompt review of the adopted technological solutions is necessary. To this end, within the scope of the methodology formation for engineering and intellectual support of railway construction technological processes, a system for monitoring and regulating work production has been developed. Intellectualization of monitoring the technological process implementation consists of the use of an expert system at the evaluation stage, built on a probabilistic inference model. The main purpose of the system is to process the monitoring results of the work progress during the railway track facility construction. The data obtained at the evaluation stage serve as the basis for predicting the implementation of the technological process. The essence of the forecast is based on the solution of general methodological issues of applying Markov processes. The regulation of the process implementation is based on the monitoring results. Adjusting the technological process to changing conditions of work provides flexibility in railway track facilities construction. This is achieved through rapid decision-making using the capabilities of an artificial neural network and subsequent adjustment of the work progress. Based on the theoretical study results, the article presents the practical aspects of monitoring and regulating the production of works in railway construction using intelligent technologies by the example of the construction of a flooded embankment section of a railway subgrade. The results presented in the article were obtained in the course of dissertation research carried out by the author.

Incorporating AI and learning analytics to build trustworthy peer assessment systems

AbstractPeer assessment has been recognised as a sustainable and scalable assessment method that promotes higher‐order learning and provides students with fast and detailed feedback on their work. Despite these benefits, some common concerns and criticisms are associated with the use of peer assessments (eg, scarcity of high‐quality feedback from peer student‐assessors and lack of accuracy in assigning a grade to the assessee) that raise questions about their trustworthiness. Consequently, many instructors and educational institutions have been anxious about incorporating peer assessment into their teaching. This paper aims to contribute to the growing literature on how AI and learning analytics may be incorporated to address some of the common concerns associated with peer assessment systems, which in turn can increase their trustworthiness and adoption. In particular, we present and evaluate our AI‐assisted and analytical approaches that aim to (1) offer guidelines and assistance to student‐assessors during individual reviews to provide better feedback, (2) integrate probabilistic and text analysis inference models to improve the accuracy of the assigned grades, (3) develop feedback on reviews strategies that enable peer assessors to review the work of each other, and (4) employ a spot‐checking mechanism to assist instructors in optimally overseeing the peer assessment process. Practitioner notesWhat is already known about this topic Engaging students in peer assessment has been demonstrated to have various benefits. However, there are some common concerns associated with employing peer assessment that raise questions about their trustworthiness as an assessment item. What this paper adds Methods and processes on how AI and learning analytics may be incorporated to address some of the common concerns associated with peer assessment systems, which in turn, can increase their trustworthiness and adoption. Implications for practice Presentation of a systematic approach for development, deployment and evaluation of AI and analytics approaches in peer assessment systems.

Extracting product competitiveness through user-generated content: A hybrid probabilistic inference model

A BERT-MDLP-Bayesian Network model (BMB) is proposed to analyze the improvement strategy of e-commerce products based on user generated content (UGC). The proposed model can be represented into four parts: clearing redundant data on the obtained UGC, extracting product attributes and word vector to generate product attributes, establishing product attribute Bayesian network corresponding to UGC, and inferring the causal relationship between product attributes. In order to verify the effectiveness of the proposed model, an amazon tablet product is used for empirical analysis. Compared with the traditional model, BMB model has better performance in product feature mining in three aspects of feature diversity, feature long tail and attribute difference. In application, the model can effectively describe the core problems of products, and provide suggestions for e-commerce to modify marketing strategies and determine the new direction of product development.

TransSounder: A Hybrid TransUNet-TransFuse Architectural Framework for Semantic Segmentation of Radar Sounder Data

Radar Sounders (RSs) are nadir-looking sensors operating in high frequency (HF) or very high frequency (VHF) bands that profile subsurface targets to retrieve miscellaneous scientific information. Due to complex electromagnetic interaction between back-scattered returns, the interpretation of RS data is challenging. The investigations of ice-sheet subsurface structures require automatic techniques to account for both the sequential spatial distribution of subsurface targets and relevant statistical properties embedded in RS signals. Automatic techniques exist for characterizing these targets either related to probabilistic inference models or convolutional neural network (CNN) deep learning methods. Unfortunately, CNN-based methods capture local spatial context and merely model the global spatial context. In contrast to CNN, the Transformer-based models are reliable architectures for capturing long-range sequence-to-sequence global spatial contextual prior. Motivated by the aforementioned fact, we propose a novel Transformer-based semantic segmentation architecture named TransSounder to effectively encode the sequential structures of the RS signals. The TransSounder was constructed on a hybrid TransUNet-TransFuse architectural framework to systematically augment the modules from TransUNet and TransFuse architectures. Experimental results obtained using the Multichannel Coherent Radar Depth Sounder (MCoRDS) dataset confirms the robustness and capability of Transformers to accurately characterize the different subsurface targets.

Spatio-Temporal Pain Estimation Network With Measuring Pseudo Heart Rate Gain

Pain is a significant indicator that shows people are suffering from an unwell experience and its automatic estimation has attracted much interest in recent years. Of late, most estimation methods are designed to capture the dynamic pain information from visual signals while a few physiological-signal based methods can provide extra potential cues to analyze the pain more accurately. However, it is still challenging to capture the physiological data from patients as it requires contact devices and patients’ cooperation. In this paper, we propose to leverage the pseudo physiological information by generating new modal data from the original visual videos and jointly estimating the pain by an end-to-end network. To extract the representations from bi-modal data, we design a spatio-temporal pain estimation network, which employs a dual-branch framework for extracting pain-aware visual and pseudo physiological features separately and fuses the features in a probabilistic way. The inherent vital sign, i.e., heart rate gain (HRG), from pseudo physiological information can be utilized as an auxiliary signal and integrated with the visual pain estimation framework. Moreover, specially-designed 3D convolution filters and attention structures are employed to extract spatio-temporal features for both branches. To use the HRG as an auxiliary way for pain estimation, we propose a probabilistic inference model by jointly considering the visual branch and physiological branch, which makes our model estimate the pain comprehensively. Experiments on two publicly-available datasets show the effectiveness of introducing the pseudo modality, and the proposed method can outperform the state-of-the-art methods.

A probabilistic Bayesian inference model to investigate injury severity in automobile crashes

Big data analytics examines millions, if not billions of records, to unmask hidden patterns, provide actionable insights and interpretable results for various domains. One area that has great potential to leverage the value of big data and analytics is the critical analysis of traffic accidents. Investigation results help in providing an in-depth understanding of the risks and provide measures to potentially prevent these risk factors hence enhancing the well-being of individuals who may experience such accidents. This study explains existing models and proposes a data science methodology in a field where probabilistic modeling makes much sense for faster, better decision-making. The main objective of this data analytics study is to identify the high-risk factors with their apparent significance to influence the probability of injury severity on automobile crashes using a geographically representative car crash dataset. To obtain reliable, accurate, and intuitive results, a multi-step probabilistic inference model based on Bayesian Belief Network— highly-acclaimed machine learning methodology—is proposed. The underlying inference model provides researchers with a causally accurate way to explore the domain (with the subject matter expert inputs) while disengaging issues related to statistical correlations and causal effects. In this study, we also used the data to create a web-based probabilistic inference simulator, a Bayesian inference decision support tool, which will be a publicly available/accessible tool, to help decision-makers better understand and to conduct what-if analysis on variable interdependencies.

Variationally Inferred Sampling through a Refined Bound.

In this work, a framework to boost the efficiency of Bayesian inference in probabilistic models is introduced by embedding a Markov chain sampler within a variational posterior approximation. We call this framework “refined variational approximation”. Its strengths are its ease of implementation and the automatic tuning of sampler parameters, leading to a faster mixing time through automatic differentiation. Several strategies to approximate evidence lower bound (ELBO) computation are also introduced. Its efficient performance is showcased experimentally using state-space models for time-series data, a variational encoder for density estimation and a conditional variational autoencoder as a deep Bayes classifier.

Probabilistic Inference Hybrid IT Value Model Using Bayesian Network

In this study, we propose probabilistic inference model on a hybrid IT value model using Bayesian Network (BN) that represents uncertain relationships between 13 variables of the model. Those variables are performance, market, innovation, IT support, core competence, capabilities, knowledge, human resources, IT development, IT resources, capital, labor, and IT spending. The relationships between variables in the model are determined using probabilistic approach, including the structure, nature, and direction of relationships. We derive a probabilistic graphical model and measure the relationships between variables. The results of this study shows that the probabilistic approach with Bayesian Network can show that capabilities and core competence are the most important variables to produce high performance output.

CSL+

Using unreliable information sources generating conflicting evidence may lead to a large uncertainty, which significantly hurts the decision making process. Recently, many approaches have been taken to integrate conflicting data from multiple sources and/or fusing conflicting opinions from different entities. To explicitly deal with uncertainty, a belief model called Subjective Logic (SL), as a variant of Dumpster-Shafer Theory, has been proposed to represent subjective opinions and to merge multiple opinions by offering a rich volume of fusing operators, which have been used to solve many opinion inference problems in trust networks. However, the operators of SL are known to be lack of scalability in inferring unknown opinions from large network data as a result of the sequential procedures of merging multiple opinions. In addition, SL does not consider deriving opinions in the presence of conflicting evidence. In this work, we propose a hybrid inference method that combines SL and Probabilistic Soft Logic (PSL), namely, Collective Subjective Plus, CSL + , which is resistible to highly conflicting evidence or a lack of evidence. PSL can reason a belief in a collective manner to deal with large-scale network data, allowing high scalability based on relationships between opinions. However, PSL does not consider an uncertainty dimension in a subjective opinion. To take benefits from both SL and PSL, we proposed a hybrid approach called CSL + for achieving high scalability and high prediction accuracy for unknown opinions with uncertainty derived from a lack of evidence and/or conflicting evidence. Through the extensive experiments on four semi-synthetic and two real-world datasets, we showed that the CSL + outperforms the state-of-the-art belief model (i.e., SL), probabilistic inference models (i.e., PSL, CSL), and deep learning model (i.e., GCN-VAE-opinion) in terms of prediction accuracy, computational complexity, and real running time.

Nested variational autoencoder for topic modelling on microtexts with word vectors

AbstractMost of the information on the Internet is represented in the form of microtexts, which are short text snippets such as news headlines or tweets. These sources of information are abundant, and mining these data could uncover meaningful insights. Topic modelling is one of the popular methods to extract knowledge from a collection of documents; however, conventional topic models such as latent Dirichlet allocation (LDA) are unable to perform well on short documents, mostly due to the scarcity of word co‐occurrence statistics embedded in the data. The objective of our research is to create a topic model that can achieve great performances on microtexts while requiring a small runtime for scalability to large datasets. To solve the lack of information of microtexts, we allow our method to take advantage of word embeddings for additional knowledge of relationships between words. For speed and scalability, we apply autoencoding variational Bayes, an algorithm that can perform efficient black‐box inference in probabilistic models. The result of our work is a novel topic model called the nested variational autoencoder, which is a distribution that takes into account word vectors and is parameterized by a neural network architecture. For optimization, the model is trained to approximate the posterior distribution of the original LDA model. Experiments show the improvements of our model on microtexts as well as its runtime advantage.

A Human-in-the-Loop Probabilistic CNN-Fuzzy Logic Framework for Accident Prediction in Vehicular Networks

The vehicle accident prediction methods are designed to improve the vehicular safety and reduce the rescue response time in the case of an accident. The existing accident prediction methods, however, do not involve Human-in-the-Loop, i.e., do not consider the emotional state of a driver to predict the likelihood of an accident. We propose a Probabilistic Convolutional Neural Network (CNN)-Fuzzy Logic framework that involves Human-in-the-Loop and takes into account the multiple input streams of sensor generated data, i.e., human emotions and traffic data. The features extracted from the CNN model are fed to our designed probabilistic graph-based inference model to determine the accident probability. The probability is then mapped with accident severity through fuzzy membership functions for accident prediction. The experiment results show the promising performance of our proposed framework, i.e., 93.1% accuracy of face expressions, 76.2% accuracy of heartbeat, and 76.9% accuracy of traffic inputs and predicts the accident likelihood with 90% accuracy. The comparison, with related works, shows that the proposed framework can predict accidents with higher probabilities.

Interactive Learning with Proactive Cognition Enhancement for Crowd Workers

Learning from crowds often performs in an active learning paradigm, aiming to improve learning performance quickly as well as to reduce labeling cost by selecting proper workers to (re)label critical instances. Previous active learning methods for learning from crowds do not have any proactive mechanism to effectively improve the reliability of workers, which prevents to obtain steadily rising learning curves. To help workers improve their reliability while performing tasks, this paper proposes a novel Interactive Learning framework with Proactive Cognitive Enhancement (ILPCE) for crowd workers. The ILPCE framework includes an interactive learning mechanism: When crowd workers perform labeling tasks in active learning, their cognitive ability to the specific domain can be enhanced through learning the exemplars selected by a psychological model-based machine teaching method. A novel probabilistic truth inference model and an interactive labeling scheme are proposed to ensure the effectiveness of the interactive learning mechanism and the performance of learning models can be simultaneously improved through a fast and low-cost way. Experimental results on three real-world learning tasks demonstrate that our ILPCE significantly outperforms five representative state-of-the-art methods.

Solving the Inverse Design Problem of Electrical Fuse With Machine Learning

Electric fuses are protection devices with a long history. They are widely used in both high voltage and low voltage power systems. The pre-arc I-t (current versus time) characteristic is an important parameter of electric fuse. The traditional method of obtaining this parameter is designing the electrical fuse by experience, then conducting experiments to obtain the I-t characteristic. However, customers request specific of I-t characters first, and then the engineers have to design electric fuses based on the requirements, this is an inverse design problem of electric fuses. There is no research in this area yet. The objective of this paper is to propose a new method to solve the inverse design problem of electrical fuse with machine learning. The method contains two steps. First, finite element analysis is used to obtain training samples for machine learning. A total of 252 samples (pre-arc current-time result) are obtained by simulations. Second, an elastic network is used to solve the probabilistic inference model by utilizing 217 samples from the total of 252. The other 35 samples are used to evaluate inverse design problem solution results. The results show that this method can accurately predict the design parameters of the restricted zone of fuse element, including the thickness (T 1 ) and radius (R 1 ) of restricted zone. The average relative errors value of T 1 and R 1 are 15.7% and 3.4%, respectively. Increasing the penalty factor and elastic error decreases the relative error. By changing the penalty error to 0.8, the average relative errors value of T 1 and R 1 are reduced by 14.1% and 1.7%, respectively. By changing the elastic error to 0.8, the average relative errors value of T 1 and R 1 are reduced by 13.1% and 2.1%, respectively. The research results provide a new solution for the electrical fuse design problem.

Medical knowledge embedding based on recursive neural network for multi-disease diagnosis.

The representation of knowledge based on first-order logic captures the richness of natural language and supports multiple probabilistic inference models. Although symbolic representation enables quantitative reasoning with statistical probability, it is difficult to utilize with machine learning models as they perform numerical operations. In contrast, knowledge embedding (i.e., high-dimensional and continuous vectors) is a feasible approach to complex reasoning that can not only retain the semantic information of knowledge, but also establish the quantifiable relationship among embeddings. In this paper, we propose a recursive neural knowledge network (RNKN), which combines medical knowledge based on first-order logic with a recursive neural network for multi-disease diagnosis. After the RNKN is efficiently trained using manually annotated Chinese Electronic Medical Records (CEMRs), diagnosis-oriented knowledge embeddings and weight matrixes are learned. The experimental results confirm that the diagnostic accuracy of the RNKN is superior to those of four machine learning models, four classical neural networks and Markov logic network. The results also demonstrate that the more explicit the evidence extracted from CEMRs, the better the performance. The RNKN gradually reveals the interpretation of knowledge embeddings as the number of training epochs increases.

Illumination animating and editing in a single picture using scene structure estimation

Editing the Illumination information in an image is a fundamental problem in image processing. In this paper, by estimating the scene structure, we propose a novel method to animate the illumination of a single input image. We first estimate the depth map of the input image by incorporating user interaction into the probabilistic inference model. Then we present a depth-aware intrinsic images decomposition method, which decomposes the image into specular shading, diffuse shading, and albedo. Combining the depth maps and shading maps, we develop a rendering-based optimization model to effectively estimate the positions and colors of multiple lights in a scene. With the estimated scene structure, including the depth map, reflectance, shading, as well as light positions and colors, we can animate and edit the illumination of the image via re-rendering the original image by changing the light configuration. Our method can effectively process images with complex shadow, multiple lights (including both point and area light sources) and specular spots. We show a variety of examples, including both indoor and outdoor environment to validate the effectiveness of the proposed method.

Optical Versus Electronic Implementation of Probabilistic Graphical Inference and Experimental Device Demonstration Using Nonlinear Photonics

The probabilistic inference model has been widely used in various areas, such as error-control coding, machine learning, speech recognition, artificial intelligence, and statistics. In this paper, we study both computation and communications power consumption of optical-based and electronic-based implementations of the probabilistic inference algorithm used in solving large scale problems. Our analysis indicates that the optical implementation provides substantial reduction for power and area compare to the electronic-based solutions as problems become large. For a network with 1 million nodes and 100 alphabet size, our proposed wavelength multiplexed all-optical implementation requires approximately 200 kilowatts (kW) of power as compared with 1.47 gigawatts (GW) and 1.7 megawatts (MW) using CPU-based and subthreshold VLSI-based systems, respectively. The optical-based solution is tolerant to shot noise and imperfections of optical modules used in the architecture as well. We also performed an all-optical experimental verification of a graphical inference as the proof of concept and have demonstrated the essential mathematical operations, multiplication, and normalization (division), in photonics operations using nonlinear bulk materials. The normalization and multiplication are shown optically through a pump-probe saturation process and a logarithm-summation-exponential (log-sum-exp) operation, respectively. We used single mode silicon waveguide and single-wall carbon nanotube (SWCNT) as nonlinear optical materials to implement logarithm and exponential operations, respectively. The SWCNT is also used as the nonlinear component in the pump-probe saturation experiment to implement the normalization function.

Deep neural network model for group activity recognition using contextual relationship

In this paper, we present contextual relationship-based learning model using deep neural network for recognizing the activities performed by a group of people in a video sequence. The proposed model comprises of the context learning using a bottom-up approach, learning from individual human actions to group level activity as well as learning from the scene information. We build deep convolutional neural network model to capture human action-pose feature for a given input video sequence. To capture group level temporal flow changes, aggregated action-pose feature of persons within the context area are fed to deep recurrent neural network, which provides spatio-temporal group descriptor. Together with this, we build a scene level convolutional neural network, to extract scene level feature which improves the performance of group activity recognition. The probabilistic inference model, as an additional layer in deep neural network, added to ensemble the models and provide a unified deep learning framework. Experimental results show the efficiency of the proposed model on standard benchmark collective activity dataset in group activity recognition. We also present the evaluated results by varying different learning parameters, optimizers, especially recurrent neural network models long short-term memory and gated recurrent unit on the benchmark collective activity dataset.

Synaptic mechanisms of interference in working memory

Information from preceding trials of cognitive tasks can bias performance in the current trial, a phenomenon referred to as interference. Subjects performing visual working memory tasks exhibit interference in their responses: the recalled target location is biased in the direction of the target presented on the previous trial. We present modeling work that develops a probabilistic inference model of this history-dependent bias, and links our probabilistic model to computations of a recurrent network wherein short-term facilitation accounts for the observed bias. Network connectivity is reshaped dynamically during each trial, generating predictions from prior trial observations. Applying timescale separation methods, we obtain a low-dimensional description of the trial-to-trial bias based on the history of target locations. Furthermore, we demonstrate task protocols for which our model with facilitation performs better than a model with static connectivity: repetitively presented targets are better retained in working memory than targets drawn from uncorrelated sequences.

Salient Region Detection via Discriminative Dictionary Learning and Joint Bayesian Inference

In past decades, saliency detection has received increasing attention from computer vision communities, for its potential usage in many vision-related tasks. However, finding representative and discriminative features to accurately locate salient regions from complex scenes remains a challenging problem. Recent research on primary visual cortex (V1) shows that vision neurons are sparsely connected to form a compact representation of natural scenes and different visual stimuli are processed separately according to their semantic importance. Inspired by the above characteristics of visual perception, in this paper we advance a novel saliency detection method via representative and discriminative dictionary learning. An assumption that salient and nonsalient information are sparsely coded under two separate dictionaries is cast on the problem and we propose to learn a compact background dictionary from the image itself for saliency estimation. Different from previous methods, our saliency cues are obtained via active learning strategies rather than artificially designed rules, and thus is more adaptive. Followed by this, a probabilistic inference model is deduced to fully excavate multisource information about the scenes for high-quality saliency map generation. This joint inference scheme takes both spatial and color space information into consideration and is proved to be quite effective in practice. Finally, to investigate the performance of the proposed model, some experiments are conducted on two benchmark data sets along with other 20 state-of-the-art saliency detection approaches. The experimental results show that our method outperforms its counterparts and can correctly detect salient regions, even when other methods fail. Besides, the usability of the proposed method in real application-based cases is verified by applying it to content-based image resizing and promising results are obtained.