Analysis of distance measures in spatial trajectory data clustering

Recent advances in location-acquisition techniques have generated massive spatial trajectory data. Recurrent Neural Networks (RNNs) are modern tools for modeling such trajectory data. After revisiting RNN-based methods for trajectory modeling, we expose two common critical drawbacks in the existing uses. First, RNNs are discrete-time models that only update the hidden states upon the arrival of new observations, which makes them an awkward fit for learning real-world trajectories with continuous-time dynamics. Second, real-world trajectories are never perfectly accurate due to unexpected sensor noise. Most RNN-based approaches are deterministic and thereby vulnerable to such noise. To tackle these challenges, we devise a novel method entitled TrajODE for more natural modeling of trajectories. It combines the continuous-time characteristic of Neural Ordinary Differential Equations (ODE) with the robustness of stochastic latent spaces. Extensive experiments on the task of trajectory classification demonstrate the superiority of our framework against the RNN counterparts.

Download Full-text

Next generation restoration metrics: Using soil eDNA bacterial community data to measure trajectories towards rehabilitation targets

10.1101/2021.08.12.456018 ◽

2021 ◽

Author(s):

Craig Liddicoat ◽

Siegy L. Krauss ◽

Andrew Bissett ◽

Ryan J. Borrett ◽

Luisa C. Ducki ◽

...

Keyword(s):

Bacterial Community ◽

Bacterial Communities ◽

High Throughput Sequencing ◽

Distance Measures ◽

Sequence Variant ◽

Trajectory Data ◽

Reference Sites ◽

Undue Influence ◽

Soil Microbiota ◽

Mining Rehabilitation

Soil microbiota are fundamentally linked to the restoration of degraded ecosystems, as they are central to important ecological functions including the support of plant communities. High throughput sequencing of environmental DNA used to characterise soil microbiota offers promise to monitor ecological progress towards reference states. In post-mining rehabilitation, successful mine closure planning requires specific, measurable, achievable, relevant and time-bound (SMART) completion criteria, such as returning ecological communities to match a target level of similarity to reference sites. We analysed patterns of surface soil bacterial community similarity to reference ('rehabilitation trajectory') data from three long-term (> 25 year) post-mining rehabilitation chronosequence case studies from south-west Western Australia. We examined the influence of different ecological distance measures, sequence grouping approaches, and eliminating rare taxa on rehabilitation trajectories and predicted recovery times. We also explored the issue of spatial autocorrelation in our rehabilitation trajectory assessments and trialled a first-pass approach for correcting its undue influence. We found considerable variation in bacterial communities among reference sites within each case study minesite, providing valuable context for setting targets and evaluating recovery. Median Bray-Curtis similarities among references within each minesite ranged from 30-36%, based on amplicon sequence variant-level data. Median predicted times for rehabilitated sites to recover to these levels ranged from around 40 to over 100 years. We discuss strengths and limitations of the different approaches and offer recommendations to improve the robustness of this assessment method. Synthesis and applications. We demonstrate a proof-of-concept, complexity-reducing application of soil eDNA sequence-based surveys of bacterial communities in restoration chronosequence studies to quantitatively assess progress towards reference communities and corresponding rehabilitation targets. Our method provides a step towards developing microbiota-based SMART metrics for measuring rehabilitation success in post-mining, and potentially other, restoration contexts. Our approach enables prediction of recovery time, explicitly including uncertainty in assessments, and assists examination of potential barriers to ecological recovery, including biologically-associated variation in soil properties.

Download Full-text

Maxmin Data Range Heuristic-based Initial Centroid Method of Partitional Clustering for Big Data Mining

International Journal of Information Retrieval Research ◽

10.4018/ijirr.289954 ◽

2022 ◽

Vol 12 (1) ◽

pp. 0-0

Keyword(s):

Big Data ◽

Data Clustering ◽

Clustering Algorithm ◽

Distance Measures ◽

Centroid Method ◽

Partitional Clustering ◽

Quantitative Measurements ◽

Centroid Distance ◽

And Performance ◽

Cluster Quality

The centroid-based clustering algorithm depends on the number of clusters, initial centroid, distance measures, and statistical approach of central tendencies. The initial centroid initialization algorithm defines convergence speed, computing efficiency, execution time, scalability, memory utilization, and performance issues for big data clustering. Nowadays various researchers have proposed the cluster initialization techniques, where some initialization techniques reduce the number of iterations with the lowest cluster quality, and some initialization techniques increase the cluster quality with high iterations. For these reasons, this study proposed the initial centroid initialization based Maxmin Data Range Heuristic (MDRH) method for K-Means (KM) clustering that reduces the execution times, iterations, and improves quality for big data clustering. The proposed MDRH method has compared against the classical KM and KM++ algorithms with four real datasets. The MDRH method has achieved better effectiveness and efficiency over RS, DB, CH, SC, IS, and CT quantitative measurements.

Download Full-text

Data Clustering: Integrating Different Distance Measures with Modified k-Means Algorithm

Advances in Intelligent and Soft Computing - Proceedings of the International Conference on Soft Computing for Problem Solving (SocProS 2011) December 20-22, 2011 ◽

10.1007/978-81-322-0491-6_63 ◽

2012 ◽

pp. 691-700 ◽

Cited By ~ 4

Author(s):

Vaishali R. Patel ◽

Rupa G. Mehta

Keyword(s):

Data Clustering ◽

Distance Measures

Download Full-text

Impact of Distance Measures on the Performance of AIS Data Clustering

Computer Systems Science and Engineering ◽

10.32604/csse.2021.014327 ◽

2021 ◽

Vol 36 (1) ◽

pp. 69-82

Author(s):

Marta Mieczyńska ◽

Ireneusz Czarnowski

Keyword(s):

Data Clustering ◽

Distance Measures

Download Full-text

A Deep Reinforcement Learning for Medical Trajectory Data Clustering Type 2 Diabetes

10.21203/rs.3.rs-888442/v1 ◽

2021 ◽

Author(s):

Arunkumar K ◽

Vasundra S

Keyword(s):

Machine Learning ◽

Deep Learning ◽

Reinforcement Learning ◽

Data Clustering ◽

Patient Treatment ◽

Learning Approaches ◽

Trajectory Data ◽

Trajectory Mining ◽

Data Investigation

Abstract Deep Reinforcement learning is incorporated in trajectory data clustering to investigate the trajectories gathered from medical information’s. Generally Trajectory mining determines the patterns in data, detects anomalies, and does informative clustering, location prediction, and classification. The main intent of Medical trajectory data clustering is identifying the trajectories with identical patterns for better patient treatment outcomes. Medical trajectory data stored in a multidimensional format which is further processed using the machine learning and deep learning architectures. Machine learning approaches employed to mine trajectory data and identifying the future treatment is a complicated task. To deal with this, the deep learning approaches in trajectory mining concentrate to eliminate the computational complexity on type 2 diabetic’s data. To overcome this problem, deep reinforcement learning for medical trajectory data clustering approach is proposed that is a combination of various strategies to flexible adapt to changes of the trajectory data. After the proposed pre-processing and feature transformation, features are clustered on basis of the weights of the model with lesser efforts and the proposed clustering plays a key role in the process of multi-attribute trajectory data investigation. The proposed deep learning methodology is more suitable for clustering the multi-attribute trajectory with fewer complexity computations than existing machine learning based methods. The experimental results also states that the results of deep reinforcement learning are promising than the other approaches with respect to precision, Recall and F Measure respectively.

Download Full-text

Interactive QOS browsing for web service selection

10.32920/ryerson.14663757 ◽

2021 ◽

Author(s):

Preethy Sambamoorthy

Keyword(s):

Web Service ◽

Data Clustering ◽

Clustering Algorithm ◽

Clustering Algorithms ◽

Service Selection ◽

Numerical Data ◽

Interval Data ◽

Distance Measures ◽

Web Service Selection

In most of the current research works on Quality of Service (QoS) based web service selection, searching is usually the dominant way to find the desired services. This approach comes with the potential problem of framing search queries properly due to requestor's lack of knowledge or vague requirement about QoS attribute values. In this thesis, we propose an interactive QoS browsing mechanism that uses the concept of clustering to present the QoS value distribution to requestors followed by finer views of service quality. By analyzing various QoS attributes, we believe that the symbolic interval data is a proper type of representation, compared with the single valued numerical data. Therefore, we use interval data clustering algorithms to implement our browsing system. We conducted experiments on simulated QoS datasets to compare the performance of using different distance measures and show the effectiveness of the interval data clustering algorithm used. The result of the experiments show that the proposed approach provides an effective, user guided QoS based service selection approach that can conceivably overcome the problems with current approaches.

Download Full-text

Interactive QOS browsing for web service selection

10.32920/ryerson.14663757.v1 ◽

2021 ◽

Author(s):

Preethy Sambamoorthy

Keyword(s):

Web Service ◽

Data Clustering ◽

Clustering Algorithm ◽

Clustering Algorithms ◽

Service Selection ◽

Numerical Data ◽

Interval Data ◽

Distance Measures ◽

Web Service Selection

In most of the current research works on Quality of Service (QoS) based web service selection, searching is usually the dominant way to find the desired services. This approach comes with the potential problem of framing search queries properly due to requestor's lack of knowledge or vague requirement about QoS attribute values. In this thesis, we propose an interactive QoS browsing mechanism that uses the concept of clustering to present the QoS value distribution to requestors followed by finer views of service quality. By analyzing various QoS attributes, we believe that the symbolic interval data is a proper type of representation, compared with the single valued numerical data. Therefore, we use interval data clustering algorithms to implement our browsing system. We conducted experiments on simulated QoS datasets to compare the performance of using different distance measures and show the effectiveness of the interval data clustering algorithm used. The result of the experiments show that the proposed approach provides an effective, user guided QoS based service selection approach that can conceivably overcome the problems with current approaches.

Download Full-text