bayesiannetworks are widely used probabilistic graphical models, whose structure is hard to learn starting from the generated data. O'Gorman et al. have proposed an algorithm to encode this task, i.e., the Bayesi...
详细信息
bayesiannetworks are widely used probabilistic graphical models, whose structure is hard to learn starting from the generated data. O'Gorman et al. have proposed an algorithm to encode this task, i.e., the bayesian network structure learning (BNSL), into a form that can be solved through quantum annealing, but they have not provided an experimental evaluation of it. In this paper, we present (i) an implementation in Python of O'Gorman's algorithm, (ii) a divide et impera approach that allows addressing BNSL problems of larger sizes in order to overcome the limitations imposed by the current architectures, and (iii) their empirical evaluation. Specifically, several problems with an increasing number of variables have been used in the experiments. The results have shown the effectiveness of O'Gorman's formulation for BNSL instances of small sizes, and the superiority of the divide et impera approach on the direct execution of O'Gorman's algorithm.
Causal machine learning (ML) algorithms recover graphical structures that tell us something about cause-and effect relationships. The causal representation provided by these algorithms enables transparency and explain...
详细信息
Causal machine learning (ML) algorithms recover graphical structures that tell us something about cause-and effect relationships. The causal representation provided by these algorithms enables transparency and explain ability, which is necessary for decision making in critical real-world problems. Yet, causal ML has had limited impact in practice compared to associational ML. This paper investigates the challenges of causal ML with application to COVID-19 UK pandemic data. We collate data from various public sources and investigate what the various structurelearning algorithms learn from these data. We explore the impact of different data formats on algorithms spanning different classes of learning, and assess the results produced by each algorithm, and groups of algorithms, in terms of graphical structure, model dimensionality, sensitivity analysis, confounding variables, predictive and interventional inference. We use these results to highlight open problems in causal structurelearning and directions for future research. To facilitate future work, we make all graphs, models, data sets, and source code publicly available online.
Bowing to the burgeoning needs of online consumers, exploitation of social media content for extrapolating buyer-centric information is gaining increasing attention of researchers and practitioners from service scienc...
详细信息
Bowing to the burgeoning needs of online consumers, exploitation of social media content for extrapolating buyer-centric information is gaining increasing attention of researchers and practitioners from service science, data analytics, machine learning and associated domains. The current paper aims to identify the structural relationship between product attributes and subsequently prioritise customer preferences with respect to these attributes while exploiting textual social media data derived from fashion blogs in Germany. A bayesian network structure learning model with the K2score maximisation objective is formulated and solved. A self-tailored metaheuristic approach that combines self-learning particle swarm optimisation (SLPSO) with the K2 algorithm (SLPSOK2) is employed to decipher the highest scored structures. The proposed approach is implemented on small, medium and large size instances consisting of 9 fashion attributes and 18 problem sets. The results obtained by SLPSOK2 are compared with the particle swarm optimisation/K2score, Genetic Algorithm/K2 score and ant colony optimisation/K2 score. Results verify that SLPSOK2 outperforms its hybrid counterparts for the tested cases in terms of computational time and solution quality. Furthermore, the study reveals that psychological satisfaction, historical revival, seasonal information and facts and figure-based reviews are major components of information in fashion blogs that influence the customers.
To solve the drawbacks of the ant colony optimization for learningbayesiannetworks (ACO-B), this paper proposes an improved algorithm based on the conditional independence test and ant colony optimization (I-ACO-B)....
详细信息
To solve the drawbacks of the ant colony optimization for learningbayesiannetworks (ACO-B), this paper proposes an improved algorithm based on the conditional independence test and ant colony optimization (I-ACO-B). First, the I-ACO-B uses order-0 independence tests to effectively restrict the space of candidate solutions, so that many unnecessary searches of ants can be avoided. And then, by combining the global score increase of a solution and local mutual information between nodes, a new heuristic function with better heuristic ability is given to induct the process of stochastic searches. The experimental results on the benchmark data sets show that the new algorithm is effective and efficient in large scale databases, and greatly enhances convergence speed compared to the original algorithm.
How to organize and manage Web services, and help users to select the atomic and a set of services with correlations to meet their functional and non-functional requirements quickly is a key problem to be solved in th...
详细信息
How to organize and manage Web services, and help users to select the atomic and a set of services with correlations to meet their functional and non-functional requirements quickly is a key problem to be solved in the era of services computing. Firstly, it uses the three-stage dependency bayesian network structure learning method to organize service clusters which realize different functions. Then it uses the maximum likelihood estimation and bayesian estimation methods to do the parameter learning, and the conditional probability table (CPT) of all the nodes can be got. This method can help users select a set of services with better function in the organized services quickly and accurately. Finally, the effectiveness of the proposed method is validated through experiments and case study.
The potential for advection to influence harmful algal bloom (HAB) spread in adjacent embayments and islands has not been investigated in the Philippines as previous studies have focused on HAB dynamics within specifi...
详细信息
The potential for advection to influence harmful algal bloom (HAB) spread in adjacent embayments and islands has not been investigated in the Philippines as previous studies have focused on HAB dynamics within specific embayments. Advection of HABs may be relevant in the Samar-Leyte region where adjacent sites are subject to recurrent blooms of Pyrodinium bahamense. We used different analyses to extract the potential role of advection in blooms in the region. First, we used bayesian and information theoretic approaches applied to historical data on shellfish bans to quantify spatial dependencies in HAB occurrences between sites. Then, to determine whether such dependencies are related to circulation patterns in the region, we analyzed connectivity using a hydrodynamic model coupled with a conservative tracer-based HAB model. The bayesiannetwork showed that in 7 out of 11 sites, the probability of a shellfish ban depended on the state of an adjacent site. Site pairs with direct dependence relations also shared relatively high similarity in HAB occurrences over time. In the modelled network, bans tend to occur sequentially, spreading from a few sites with relatively high probabilities for ban events. A subset of sites (sources) were found to be informative of future HAB event probabilities in other sites (destinations) over time lags that are generally longer the farther the destination. Modelled surface advection showed high connectivity strengths between sources and destinations associated with circulation features, e.g., an anticyclonic current in Leyte, wind-driven coastal current in western Samar, and tidally-driven flow in the shallow embayments in southwest Samar. High connectivities were correlated with direct dependence relations in the bayesiannetwork. Connectivity explained up to about 1/3 of the variance in statistical dependencies between ban signals. Our results show that Paralytic Shellfish Toxin events within this region can be due to advection of bl
bayesiannetwork is one of the most efficient and reliable method in data mining, and bayesian network structure learning is a key link in the process of bayesiannetwork research. Aiming at the problem of the classic...
详细信息
ISBN:
(纸本)9781467377249
bayesiannetwork is one of the most efficient and reliable method in data mining, and bayesian network structure learning is a key link in the process of bayesiannetwork research. Aiming at the problem of the classic Hill-Climbing algorithm is easy to fall into local optimum and low in efficiency, establishing the Most Weight Supported Tree by calculating the mutual information, and combining the Most Weight Supported Tree and the simplified Hill-Climbing algorithm, proposes a new improved bayesian network structure learning algorithm. Comparing with the classic Hill-Climbing algorithm and K2 algorithm, the simulation experiments shown that the improved algorithm not only can obtain a high accuracy rate model, but improve the efficiency of building model. Based on the improved algorithm and combined with JiDong cement's cement rotary kiln operating data, we can establish the fault diagnosis model of cement rotary kiln and realize a precise and rapid fault diagnosis.
暂无评论