Driftage: a multi-agent system framework for concept drift detection.
Ontology highlight
ABSTRACT: Background
The amount of data and behavior changes in society happens at a swift pace in this interconnected world. Consequently, machine learning algorithms lose accuracy because they do not know these new patterns. This change in the data pattern is known as concept drift. There exist many approaches for dealing with these drifts. Usually, these methods are costly to implement because they require (i) knowledge of drift detection algorithms, (ii) software engineering strategies, and (iii) continuous maintenance concerning new drifts.Results
This article proposes to create Driftage: a new framework using multi-agent systems to simplify the implementation of concept drift detectors considerably and divide concept drift detection responsibilities between agents, enhancing explainability of each part of drift detection. As a case study, we illustrate our strategy using a muscle activity monitor of electromyography. We show a reduction in the number of false-positive drifts detected, improving detection interpretability, and enabling concept drift detectors' interactivity with other knowledge bases.Conclusion
We conclude that using Driftage, arises a new paradigm to implement concept drift algorithms with multi-agent architecture that contributes to split drift detection responsability, algorithms interpretability and more dynamic algorithms adaptation.
SUBMITTER: Vieira DM
PROVIDER: S-EPMC8168350 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Introduction: Clinical guidelines recommend best care pathways for many clinical conditions. When significant data from clinical trials become available, new clinical guidelines are published, formalizing the advances in the field. When we consider clinical guidelines as processes of care, we can use process mining techniques to discover how innovations diffuse into healthcare and how they modify the execution of clinical processes. Methods: We conducted a study to assess the changes in process execution patterns after a 2013 update of an acute ischemic stroke (AIS) guideline. We used MIMIC-IV as the data source, including patients from 2008 until 2019 hospitalized with an AIS and applied drift detection methods to measure changes in the therapeutic process. We performed statistical tests to determine whether the underlying distribution of events reflects the changes in the guidelines post-2013. Results: Ischemic stroke patients show few significant changes in clinical practices, despite an update in guidelines. The positive control group of aortic valve replacement patients shows a significant change in clinical practices surrounding this procedure. Conclusions: This study demonstrates the use of drift detection methods as a novel method to study the diffusion of innovations in healthcare settings from a process perspective.
| S-EPMC10148346 | biostudies-literature
Project description:Wireless sensor networks (WSNs) have been widely investigated in recent years. One of the fundamental issues in WSNs is packet routing, because in many application domains, packets have to be routed from source nodes to destination nodes as soon and as energy efficiently as possible. To address this issue, a large number of routing approaches have been proposed. Although every existing routing approach has advantages, they also have some disadvantages. In this paper, a multi-agent framework is proposed that can assist existing routing approaches to improve their routing performance. This framework enables each sensor node to build a cooperative neighbour set based on past routing experience. Such cooperative neighbours, in turn, can help the sensor to effectively relay packets in the future. This framework is independent of existing routing approaches and can be used to assist many existing routing approaches. Simulation results demonstrate the good performance of this framework in terms of four metrics: average delivery latency, successful delivery ratio, number of live nodes and total sensing coverage.
| S-EPMC4481995 | biostudies-other
Project description:A financial market is a platform to produce data streams continuously and around 1. 145 Trillion MB of data per day. Estimation and the analysis of unknown or dynamic behaviors of these systems is one the challenging tasks. Analysis of these systems is very much essential to strengthen the environmental parameters to stabilize society activities. This can elevate the living style of society to the next level. In this connection, the proposed paper is trying to accommodate the financial data stream using the sliding window approach and random forest algorithm to provide a solution to handle concept drift in the financial market to stabilize the behavior of the system through drift estimation. The proposed approach provides promising results in terms of accuracy in detecting concept drift over the state of existing drift detection methods like one class drifts detection (OCDD), Adaptive Windowing ADWIN), and the Page-Hinckley test.
| S-EPMC9471369 | biostudies-literature
Project description:This study aims to investigate the problem of concept drift in cloud computing and emphasizes the importance of early detection for enabling optimum resource utilization and offering an effective solution. The analysis includes synthetic and real-world cloud datasets, stressing the need for appropriate drift detectors tailored to the cloud domain. A modified version of Long Short-Term Memory (LSTM) called the LSTM Drift Detector (LSTMDD) is proposed and compared with other top drift detection techniques using prediction error as the primary evaluation metric. LSTMDD is optimized to improve performance in detecting anomalies in non-Gaussian distributed cloud environments. The experiments show that LSTMDD outperforms other methods for gradual and sudden drift in the cloud domain. The findings suggest that machine learning techniques such as LSTMDD could be a promising approach to addressing the problem of concept drift in cloud computing, leading to more efficient resource allocation and improved performance.
| S-EPMC10909158 | biostudies-literature
Project description:A two-stage decomposition approach based on a novel multi-agent system (MAS) is proposed for the distributed resource constrained multi-project scheduling problem (DRCMPSP). In stage one, from the point of view of each local project manager, a forward-backward hybrid genetic algorithm (FBHGA) is developed to generate an initial local schedule with the objective of minimizing individual project makespan. In stage two, from the global perspective of project management office, a sequential game-based negotiation mechanism is employed to eliminate global resource conflicts with the objective of minimizing total tardiness cost (TTC). The proposed approach is tested on 140 benchmark problem instances. According to the computational results, high-quality local project schedules can be obtained by FBHGA in stage one. Furthermore, it is observed that our method is capable of dealing with various complex multi-project instances under different degrees of resource conflicts in reasonable CPU running time. Compared to the existing decentralized methods for DRCMPSP, the proposed approach with sequential game-based negotiation mechanism shows the superiority in producing multi-project schedules with lower TTC, especially for large-size and strong conflicting instances.
| S-EPMC6177185 | biostudies-literature
Project description:MotivationDue to the complexity and heterogeneity of multicellular biological systems, mathematical models that take into account cell signalling, cell population behaviour and the extracellular environment are particularly helpful. We present PhysiBoSS, an open source software which combines intracellular signalling using Boolean modelling (MaBoSS) and multicellular behaviour using agent-based modelling (PhysiCell).ResultsPhysiBoSS provides a flexible and computationally efficient framework to explore the effect of environmental and genetic alterations of individual cells at the population level, bridging the critical gap from single-cell genotype to single-cell phenotype and emergent multicellular behaviour. PhysiBoSS thus becomes very useful when studying heterogeneous population response to treatment, mutation effects, different modes of invasion or isomorphic morphogenesis events. To concretely illustrate a potential use of PhysiBoSS, we studied heterogeneous cell fate decisions in response to TNF treatment. We explored the effect of different treatments and the behaviour of several resistant mutants. We highlighted the importance of spatial information on the population dynamics by considering the effect of competition for resources like oxygen.Availability and implementationPhysiBoSS is freely available on GitHub (https://github.com/sysbio-curie/PhysiBoSS), with a Docker image (https://hub.docker.com/r/gletort/physiboss/). It is distributed as open source under the BSD 3-clause license.Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC6449758 | biostudies-literature
Project description:Remote sensing using satellites and unmanned aerial vehicles (UAVs) has become an important tool for wetland delimitation and saturation assessment since they enable patterns identification and wetland saturation data collection in an agile and optimum way. However, their deployment and operative costs limit their implementation in harsh environments, such as the ones presented in the high Andean wetlands. In this context, this work presents a framework to monitor cost-effectively high Andean wetlands using a multi-agent approach based on: field testing, UAV orthomosaics, and satellite imagery. The method developed comprises two stages: i) definition of the monitoring agent (field testing, satellite, UAV) and ii) image processing. For these stages, semi-empirical and statistical models, which were developed in previous works are incorporated in an open-source framework to tailor each monitoring approach accordingly to the seasonality of a representative Andean wetland. The application of the method and its results highlight the suitability of using visual spectrum low-cost remote sensing approach to compute wetlands saturation percentage. In addition, the methodology proposed allowed the development of a temporal monitoring scheme, where the viability of each monitoring agent is examined. In order to validate the method, field data and multispectral imagery were employed using as case of study the Pugllohuma wetland located in the Antisana Reserve. Thus, the main contribution of this work lies in establishing a technified monitoring framework for the Ecuadorian high Andean wetlands, which can be scaled up and extrapolated to other wetlands with similar harsh environmental conditions, helping to their management and protection policies decision-making.
| S-EPMC8968649 | biostudies-literature
Project description:Proteins from thermophiles are generally more thermostable than their mesophilic homologs, but little is known about the evolutionary process driving these differences. Here we attempt to understand how the diverse thermostabilities of bacterial ribonuclease H1 (RNH) proteins evolved. RNH proteins from Thermus thermophilus (ttRNH) and Escherichia coli (ecRNH) share similar structures but differ in melting temperature (T(m)) by 20 °C. ttRNH's greater stability is caused in part by the presence of residual structure in the unfolded state, which results in a low heat capacity of unfolding (ΔCp) relative to ecRNH. We first characterized RNH proteins from a variety of extant bacteria and found that Tm correlates with the species' growth temperatures, consistent with environmental selection for stability. We then used ancestral sequence reconstruction to statistically infer evolutionary intermediates along lineages leading to ecRNH and ttRNH from their common ancestor, which existed approximately 3 billion years ago. Finally, we synthesized and experimentally characterized these intermediates. The shared ancestor has a melting temperature between those of ttRNH and ecRNH; the T(m)s of intermediate ancestors along the ttRNH lineage increased gradually over time, while the ecRNH lineage exhibited an abrupt drop in Tm followed by relatively little change. To determine whether the underlying mechanisms for thermostability correlate with the changes in T(m), we measured the thermodynamic basis for stabilization--ΔCp and other thermodynamic parameters--for each of the ancestors. We observed that, while the T(m) changes smoothly, the mechanistic basis for stability fluctuates over evolutionary time. Thus, even while overall stability appears to be strongly driven by selection, the proteins explored a wide variety of mechanisms of stabilization, a phenomenon we call "thermodynamic system drift." This suggests that even on lineages with strong selection to increase stability, proteins have wide latitude to explore sequence space, generating biophysical diversity and potentially opening new evolutionary pathways.
| S-EPMC4227636 | biostudies-literature
Project description:In recent years, maintenance work on public transport routes has drastically decreased in many countries due to difficult economic situations. The various studies that have been conducted by groups of drivers and groups related to road safety concluded that accidents are increasing due to the poor conditions of road surfaces, even affecting the condition of vehicles through costly breakdowns. Currently, the processes of detecting any type of damage to a road are carried out manually or are based on the use of a road vehicle, which incurs a high labor cost. To solve this problem, many research centers are investigating image processing techniques to identify poor-condition road areas using deep learning algorithms. The main objective of this work is to design of a distributed platform that allows the detection of damage to transport routes using drones and to provide the results of the most important classifiers. A case study is presented using a multi-agent system based on PANGEA that coordinates the different parts of the architecture using techniques based on ubiquitous computing. The results obtained by means of the customization of the You Only Look Once (YOLO) v4 classifier are promising, reaching an accuracy of more than 95%. The images used have been published in a dataset for use by the scientific community.
| S-EPMC7663208 | biostudies-literature
Project description:Retinal images play a pivotal contribution to the diagnosis of various ocular conditions by ophthalmologists. Extensive research was conducted to enable early detection and timely treatment using deep learning algorithms for retinal fundus images. Quick diagnosis and treatment planning can be facilitated by deep learning models' ability to process images rapidly and deliver outcomes instantly. Our research aims to provide a non-invasive method for early detection and timely eye disease treatment using a Convolutional Neural Network (CNN). We used a dataset Retinal Fundus Multi-disease Image Dataset (RFMiD), which contains various categories of fundus images representing different eye diseases, including Media Haze (MH), Optic Disc Cupping (ODC), Diabetic Retinopathy (DR), and healthy images (WNL). Several pre-processing techniques were applied to improve the model's performance, such as data augmentation, cropping, resizing, dataset splitting, converting images to arrays, and one-hot encoding. CNNs have extracted extract pertinent features from the input color fundus images. These extracted features are employed to make predictive diagnostic decisions. In this article three CNN models were used to perform experiments. The model's performance is assessed utilizing statistical metrics such as accuracy, F1 score, recall, and precision. Based on the results, the developed framework demonstrates promising performance with accuracy rates of up to 89.81% for validation and 88.72% for testing using 12-layer CNN after Data Augmentation. The accuracy rate obtained from 20-layer CNN is 90.34% for validation and 89.59% for testing with Augmented data. The accuracy obtained from 20-layer CNN is greater but this model shows overfitting. These accuracy rates suggested that the deep learning model has learned to distinguish between different eye disease categories and healthy images effectively. This study's contribution lies in providing a reliable and efficient diagnostic system for the simultaneous detection of multiple eye diseases through the analysis of color fundus images.
| S-EPMC11271906 | biostudies-literature