Drift Problem Research Articles

AIMED: An automatic and incremental approach for business process model repair under concept drift

Real-life business processes may change over time in response to new business requirements, market changes, new policies or regulations, etc., which is called concept drift in the data mining area. How to identify and deal with the concept drift problem is a significant challenge in business process mining. Currently, the research mainly focuses on the problems of concept drift detection. However, the proposed approaches do not intuitively explain how the process changes. In this paper, we provide an integrated framework whose core technique is an Automatic and Incremental approach for business process Model rEpair under concept Drift (AIMED). AIMED streamlines the functions of concept drift detection, localization and model repair. More specifically, when concept drift is detected, it updates the process model automatically by precisely localizing the sub-structure of the process model that concept drift affects and updating this sub-structure accordingly. Concept drift is explained intuitively by presenting the repaired model. In particular, AIMED can resist the noise that greatly affects the performance of current concept drift detection and process model repair techniques. The properties of AIMED are theoretically proven. We also conduct extensive experiments on synthetic logs as well as real-life logs. The experiment results show that AIMED outperforms the state-of-the-art methods in both concept drift detection and process model repair and works well even when there is noise in logs.

ADTCD: An Adaptive Anomaly Detection Approach Toward Concept Drift in IoT

The data collected by sensors is streaming data in the Internet of Things(IoT). Although existing deep learningbased anomaly detection methods generally perform well on static data, they struggle to respond timely to streaming data after distribution changes. However, streaming data suffers from conceptual drift due to the highly dynamic nature of IoT. In network security, concept drift-oriented anomaly detection is a crucial task, because it can adjust the model to adapt to the latest data, and detect attacks in time. Existing streaming anomaly detection methods are confronted with some challenges, including of the latency of model updates, the uneven importance of new data, and the self-poisoning due to model self-updates. To tackle the above challenges, we propose a knowledge distillation based adaptive anomaly detection model towards concept-drift, ADTCD. ADTCD transfers the knowledge of the teacher model to the student model and only updates the student model to reduce the delay. We construct an algorithm of dynamically adjusting model parameters, which dynamically adjusts model weights through local inference on new samples, in order to improving the models responsiveness to new distribution data, meanwhile solving the problem of uneven importance of new data. In additions, we adopt a one-class support vector-based outlier removal method to tackle the self-poisoning problem. In comprehensive experiments on seven high-dimensional datasets, ADTCD achieves an AUC improvement of 12.46% compared to the state-of-the-art streaming anomaly detection methods. Our future direction will focus on exploring the concept drift problem using methods beyond autoencoders.

Improved Federal Test Procedure (FTP75) driving cycle performance for PMSM‐fed hybrid electric vehicles using artificial neural network

SummaryThe sensorless speed control of permanent magnet synchronous motor (PMSM) is gaining popularity in hybrid electric vehicle (HEV) applications leading to its enhanced safety, reliability, and cost savings. Speed control using vector control for PMSM‐fed HEV requires the speed encoder. When the speed sensor information fails, the inverter must ensure power delivery to the PMSM continuously by estimating the speed; this mode of operation is referred as limp‐home mode in HEV. In this paper, a speed sensorless scheme has been proposed for PMSM‐based HEV during limp‐home mode operation. This paper presents a model reference adaptive system (MRAS) speed estimator based on an adaptive neural network controller (NNC) for speed estimation of PMSM. In the HEV application, in case of speed/position encoder failure, the speed of the PMSM can be estimated by stator flux using stator current.The proposed method employs stator currents in the reference model to eliminate the DC drift problem. Furthermore, the NNC is employed in the adaptation mechanism to improve the Federal Test Procedure (FTP75) driving cycle performance. The performance of the proposed control scheme has been validated with dSPACE 1104 R & D rapid development controller using vector control for the PMSM during variable speed and torque, including the zero‐speed applications

Predicting cancer outcomes from whole slide images via hybrid supervision learning

Collaboratively leveraging limited pixel-level segmentation annotations and large-scale slide-level classification labels in hybrid supervision learning can significantly enhance model performance. However, the direct application of this approach within computational pathology presents challenges as end-to-end classification models grapple with processing high-resolution whole slide images (WSIs). An alternative approach is to use patch-based models with pixel-level pseudo-labels, but these models can be susceptible to the cumulative effects of noisy labels, leading to convergence and drift problems during iterative training. To surmount these hurdles, we propose a hybrid supervision learning framework tailored for pathological image classification. Our method employs coarse classification labels to optimize pixel-level pseudo-labels and incorporates a comprehensive strategy to diminish false negatives and positives throughout the segmentation process. This framework holistically integrates the supervised information derived from segmentation and classification procedures, and is applicable to the general classification of high-resolution images, thereby boosting both specificity and sensitivity. We assess our proposed method’s effectiveness using one publicly accessible dataset and two proprietary datasets, collectively constituting over 10,000 pathological images of various disease types such as gastric, cervical, and breast cancer. Our experimental results reveal a 100% sensitivity rate in slide-level classification tasks, simultaneously reducing the false-positive rate to a mere third of the state-of-the-art. In conclusion, this paper presents a potent instrument for the precise and efficient classification of high-resolution pathological images, with promising results showcased across a wide array of datasets and disease types. Code is available at https://github.com/JarveeLee/HybridSupervisionLearning_Pathology.

A High-Precision Quartz Resonant Ultra-High Pressure Sensor with Integrated Pressure Conversion Structure.

A quartz resonant pressure sensor is proposed for high-precision measurement of ultra-high pressure. The resonant unit realizes a push-pull differential layout, which restrains the common-mode interference factor, and the resonator is only subject to axial force. The pressure conversion unit is made in an integrated manner, avoiding output drift problems caused by residual stress and small gaps during assembly, welding, and other processes in sensor preparation. Theoretical and simulation analysis was conducted on the overall design scheme of the sensor in this paper, verifying the feasibility. Sensor prototypes were created and performance experiments were conducted. The experimental results show that the sensitivity of the ultra-high pressure sensor is 46.32 Hz/MPa at room temperature within the pressure range of 120 MPa, and the comprehensive accuracy is 0.0266%. The comprehensive accuracy of the sensor is better than 0.0288% FS in the full temperature range environment. This proves that the sensor scheme is suitable for high-precision and high-stability detection of ultra-high pressure, providing new solutions in special pressure measurement fields such as deep-sea and oil exploration.

Using N-BEATS ensembles to predict automated guided vehicle deviation

AbstractA novel AGV (Automated Guided Vehicle) control architecture has recently been proposed where the AGV is controlled remotely by a virtual Programmable Logic Controller (PLC), which is deployed on a Multi-access Edge Computing (MEC) platform and connected to the AGV via a radio link in a 5G network. In this scenario, we leverage advanced deep learning techniques based on ensembles of N-BEATS (state-of-the-art in time-series forecasting) to build predictive models that can anticipate the deviation of the AGV’s trajectory even when network perturbations appear. Therefore, corrective maneuvers, such as stopping the AGV, can be performed in advance to avoid potentially harmful situations. The main contribution of this work is an innovative application of the N-BEATS architecture for AGV deviation prediction using sequence-to-sequence modeling. This novel approach allows for a flexible adaptation of the forecast horizon to the AGV operator’s current needs, without the need for model retraining or sacrificing performance. As a second contribution, we extend the N-BEATS architecture to incorporate relevant information from exogenous variables alongside endogenous variables. This joint consideration enables more accurate predictions and enhances the model’s overall performance. The proposed solution was thoroughly evaluated through realistic scenarios in a real factory environment with 5G connectivity and compared against main representatives of deep learning architectures (LSTM), machine learning techniques (Random Forest), and statistical methods (ARIMA) for time-series forecasting. We demonstrate that the deviation of AGVs can be effectively detected by using ensembles of our extended N-BEATS architecture that clearly outperform the other methods. Finally, a careful analysis of a real-time deployment of our solution was conducted, including retraining scenarios that could be triggered by the appearance of data drift problems.

Learning Data Streams With Changing Distributions and Temporal Dependency.

In a data stream, concept drift refers to unpredictable distribution changes over time, which violates the identical-distribution assumption required by conventional machine learning methods. Current concept drift adaptation techniques mostly focus on a data stream with changing distributions. However, since each variable of a data stream is a time series, these variables normally have temporal dependency problems in the real world. How to solve concept drift and temporal dependency problems at the same time is rarely discussed in the concept-drift literature. To solve this situation, this article proves and validates that the testing error decreases faster if a predictor is trained on a temporally reconstructed space when drift occurs. Based on this theory, a novel drift adaptation regression (DAR) framework is designed to predict the label variable for data streams with concept drift and temporal dependency. A new statistic called local drift degree (LDD+) is proposed and used as a drift adaptation technique in the DAR framework to discard outdated instances in a timely way, thereby guaranteeing that the most relevant instances will be selected during the training process. The performance of DAR is demonstrated by a set of experimental evaluations on both synthetic data and real-world data streams.

Deep Incremental Hashing for Semantic Image Retrieval With Concept Drift

Hashing methods are widely used for content-based image retrieval due to their attractive time and space efficiencies. Several dynamic hashing methods have been proposed for image retrieval tasks in non-stationary environments. However, concept drift problems in non-stationary environment are seldomly considered which lead to significant deterioration of performance. Therefore, we propose Deep Incremental Hashing (DIH). For the learning part, similarity-preserving object codes of each newly arriving data chunk are computed using the product of its label matrix and a random Gaussian matrix generated offline. A point-wise loss function is then devised to guide the learning of a deep hash neural network. To retain the learned knowledge of former chunks, a weighting-based method is utilized to combine different hash tables trained at different time steps to form a multi-table hashing system. Experimental results on 13 simulated concept drift environments show that DIH adapts to non-stationary data environments well and yields better retrieval performance than existing dynamic hashing methods.

Design of Real-time Test System for Servo Stabilization System Based on Matlab/xPC

For the barrel and static drift problem in the working process of the servo-stabilization system of an unmanned firing module, a simulation test system is built by using the Matlab/xPC real-time platform. The CAN communication card is driven by the Simulink S-function, and communication between the test system and the servo stabilization system is realized. The fast Haar wavelet threshold processing method is used to de-noise the gyro speed signal, and the de-noising effect of the signal processor is simulated and verified in the physical system. The results show that the servo stability test system has strong real-time performance, and the performance of the Haar wavelet de-noising algorithm is good. The system has a broad application value.

A linear adjustment-based approach to posterior drift in transfer learning.

We present new models and methods for the posterior drift problem where the regression function in the target domain is modelled as a linear adjustment, on an appropriate scale, of that in the source domain, and study the theoretical properties of our proposed estimators in the binary classification problem. The core idea of our model inherits the simplicity and the usefulness of generalized linear models and accelerated failure time models from the classical statistics literature. Our approach is shown to be flexible and applicable in a variety of statistical settings, and can be adopted for transfer learning problems in various domains including epidemiology, genetics and biomedicine. As concrete applications, we illustrate the power of our approach (i) through mortality prediction for British Asians by borrowing strength from similar data from the larger pool of British Caucasians, using the UK Biobank data, and (ii) in overcoming a spurious correlation present in the source domain of the Waterbirds dataset.

Benchmarking EEG-based Cross-dataset Driver Drowsiness Recognition with Deep Transfer Learning.

It usually takes a long time to collect data for calibration when using electroencephalography (EEG) for driver drowsiness monitoring. Cross-dataset recognition is desirable since it can significantly save the calibration time when an existing dataset is used. However, the recognition accuracy is affected by the distribution drift problem caused by different experimental environments when building different datasets. In order to solve the problem, we propose a deep transfer learning model named Entropy-Driven Joint Adaptation Network (EDJAN), which can learn useful information from source and target domains simultaneously. An entropy-driven loss function is used to promote clustering of target-domain representations and an individual-level domain adaptation technique is proposed to alleviate the distribution discrepancy problem of test subjects. We use two public driving datasets SEEG-VIG and SADT to test the model on the cross-dataset setting. The proposed model achieved an accuracy of 83.3% when SADT is used as source domain and SEED-VIG is used as target domain and 76.7% accuracy on the reverse setting, which is higher than the other SOTA methods. The results are further analyzed with both global and local interpretation methods. Our work illuminates a promising direction of using EEG for calibration-free driver drowsiness recognition.

Capturing geometric structure change through deformation aware correlation

Capturing the inner structure change of the target is crucial for a siamese tracker confronted with severe deformation or occlusion. The conventional siamese tracker utilizes a strict grid-to-grid correlation to pass the template information towards the current candidate, which may suffer severe mismatch compared to the initial frame caused by undesirable conditions such as rotation and deformation. Thus such mismatch would lead to insufficient information passed from the template and eventually end with a poor performance in classification and regression branches. The inherent CNN structure with the restricted grid-to-grid matching mechanism during correlation is not able to capture the corresponding change of the target structure. Motivated by DCNv2, we propose a novel deformation aware correlation to replace the widely-used depthwise-correlation. An offset-prediction module is proposed based on the similarity matrix to compensate for the mismatch of the grid assignment between template and candidate. Then we utilize an importance bilinear sampling algorithm to assess the similarity of the matching pair, alleviating the occlusion drift problem. Finally, a modulated deformable correlation is fed into the aforementioned input and realizes a better information modulation. We replace the correlation operator on SiamRPN++ with our deformation aware correlation operator and obtain our tracker, SiamDAC. Massive experiments on OTB100, UAV123, VOT2019, LaSOT and DTB70 validate the effectiveness of our proposed operator, especially confronted with severe deformation or occlusion.

Transductive semantic knowledge graph propagation for zero-shot learning

The domain shift is a very common phenomenon in Zero-shot Learning (ZSL), because the data distribution between the source and target domain might not match well in real scenarios. This paper focuses on dealing with the domain drift problem in ZSL based on the semantic knowledge graph propagation. Our method consists of visual feature extraction (VFE) module, semantic feature extraction module (SFE) and feature mapping module (FM). In VFE module, the high-level visual features are extracted under Convolutional Neural Network (CNN) framework, where unseen samples participate in model training and stay away from seen ones under the premise of distinguishability. SFE module relies on the Graph Convolutional Network (GCN) framework and focuses on the transmission of semantic embeddings. A modified message aggregation and transformation strategy is proposed to effectively relieve the information smoothing phenomenon caused by the increase of GCN layers. In the recognition stage, AutoEncoder framework achieves a two-way visual and semantic interaction to further minimize the domain drift phenomenon. Simulation results on ImageNet dataset have demonstrated the performance of the proposed method.

Structural damage identification with unknown external inputs based on the sparse constraint

The strength and integrity of a structure are determined not only by the quality of its materials, but also by the health of its various components. In some cases, damaged cracks inside the structure will be generated from excessive loads, which may cause catastrophic failure of structures if undetected for a long period. However, at early stages, most of the damages are minor, and therefore difficult to detect by only visual inspection. Thus, in this paper, a damage estimation algorithm based on an unscented Kalman filter (UKF) is proposed, which can identify and locate the damage parameters in real time using a limited number of sensors. Meanwhile, using this algorithm, joint force-damage estimation can be achieved, which is very applicable to the structural system with unknown external inputs. On the other hand, for most structures, the distribution of damage parameters in the space domain is sparse. Therefore, the sparsity of the damage parameter vector is introduced to UKF as an l 1-norm constraint by the pseudo measurement (PM) technique. Thus, unconstrained optimization of the damage parameter estimation is transformed into an l1 -norm constrained optimization problem. With such improvement, the process of damage parameter estimation converges faster, and the false damage parameters can be effectively restrained. Moreover, to solve the force drift problem during force identification if only acceleration data is used, the sparse constraint of the force vector is also introduced to the UKF framework by the PM technique. Finally, the performance of the proposed algorithm is validated by two case studies, including numerical simulations of a ten-story shear building and experiments of a three-story shear structure. The results indicate that the proposed algorithm can accurately identify the damage, and successfully resolve the common force drift problem.

An Improved Method for Accurate Radiation Measurement Based on Dark Output Noise Drift Compensation.

This paper verified through experiments that change in ambient temperature are the main cause of dark output noise drift. Additionally, the impact of dark output noise drift in fiber optic spectrometers on emissivity measurements has been investigated in this work. Based on an improved fiber optic spectrometer, two methods were proposed for characterizing and correcting the dark output noise offset in fiber optic spectrometers: the mean correction scheme and the linear fitting correction scheme. Compared to the mean correction scheme, the linear fitting correction scheme is more effective in solving the problem of dark output noise drift. When the wavelength is greater than 1600 nm, the calibration relative error of silicon carbide (SIC) emissivity is less than 0.8% by the mean correction scheme, while the calibration relative error of silicon carbide emissivity is less than 0.62% by the linear fitting correction scheme. This work solves the problem of dark output noise drift in prolonged measurement based on fiber optic spectrometers, improving the accuracy and reliability of emissivity and quantitative radiation measurement.

Baseline correction using morphological and iterative local extremum (MILE)

Baseline drift problem usually exists in the spectral analysis due to the detector nonlinearities and temperature variations. Therefore, the baseline must be corrected before accurate data analysis. So, we proposed a novel and high-precise algorithm, baseline estimation using morphological and iterative local extremum (MILE). We first obtained all local extrema of the measured spectrum by derivation. A coarse baseline is then estimated by the PCHIP interpolation method. Subsequently, find the local extrema again of coarse baseline and update by their adjacent data. The update processing iterates multiple times. Finally, the spectrum is corrected by subtracting the estimated baseline. In order to evaluate the performance of our algorithm, we corrected the baseline drift of various spectra, including chromatograms and Fourier Transform Spectrometer spectra. All results achieve a high precision, proving that our algorithm has better accuracy and robustness.

Spatial-temporal hypergraph convolutional network for traffic forecasting.

Accurate traffic forecasting plays a critical role in the construction of intelligent transportation systems. However, due to the across road-network isomorphism in the spatial dimension and the periodic drift in the temporal dimension, existing traffic forecasting methods cannot satisfy the intricate spatial-temporal characteristics well. In this article, a spatial-temporal hypergraph convolutional network for traffic forecasting (ST-HCN) is proposed to tackle the problems mentioned above. Specifically, the proposed framework applies the K-means clustering algorithm and the connection characteristics of the physical road network itself to unify the local correlation and across road-network isomorphism. Then, a dual-channel hypergraph convolution to capture high-order spatial relationships in traffic data is established. Furthermore, the proposed framework utilizes a long short-term memory network with a convolution module (ConvLSTM) to deal with the periodic drift problem. Finally, the experiments in the real world demonstrate that the proposed framework outperforms the state-of-the-art baselines.

HELOP: Multi-target tracking based on heuristic empirical learning algorithm and occlusion processing

Multi-target tracking is one of the important fields in computer vision, which aims to solve the problem of matching and correlating targets between adjacent frames. In this paper, we propose a fine-grained track recoverable (FGTR) matching strategy and a heuristic empirical learning (HEL) algorithm. The FGTR matching strategy divides the detected targets into two different sets according to the distance between them, and adopts different matching strategies respectively, in order to reduce false matching, we evaluated the trust degree of the target’s appearance feature information and location feature information, adjusted the proportion of the two reasonably, and improved the accuracy of target matching. In order to solve the problem of trajectory drift caused by the cumulative increase of Kalman filter error during the occlusion process, the HEL algorithm predicts the position information of the target in the next few frames based on the effective information of other previous target trajectories and the motion characteristics of related targets. Make the predicted trajectory closer to the real trajectory. Our proposed method is tested on MOT16 and MOT17, and the experimental results verify the effectiveness of each module, which can effectively solve the occlusion problem and make the tracking more accurate and stable.

In-situ calibration method for thermocouples in accelerating rate calorimeter based on multiple fixed-points and Joule heat

To solve the problem of temperature drift of thermocouples in accelerating rate calorimeter caused by long-term operation, an in-situ calibration method based on multiple fixed-points and Joule heat is proposed in this article. Firstly, the heat transfer model of the calorimeter is established, and the validity of the method is verified by numerical simulation. Secondly, a multiple fixed-points graphite calibrator and a cylindrical electronic resistance element are designed. Finally, the in-situ calibration is carried out. The calibration results show that the maximum permissible measurement error of the sample thermocouple after calibration is better than 0.220 °C and that the bias of consistency in-situ calibration method is smaller than 0.110 °C. In addition, a di‑tert‑butyl peroxide in toluene solution with a mass percent of 20% is selected as the experimental sample. The sample experiment results show that the kinetic and thermodynamic parameters are closer to the reference values after thermocouples calibration.

Visual-haptic feedback for ROV subsea navigation control

Subsea engineering operations, including subsea inspection, installation and maintenance, heavily rely on the seamless interaction between remotely operated vehicles (ROV) and human operators. Subsea ROV control has always been a great challenge to human operators due to the dynamics of the subsea environment such as uncertainty of turbulence, affected visibility and interference with subsea ecosystems. Compared to engineering work on the ground, subsea engineering tasks are usually finished in a limited vision field with drift and localization problems. However, the traditional ROV feedback system, the live video streaming, cannot provide direct and intuitive perceptions of remote ROV workspace as well as a clear indication of the workspace uncertainties. Therefore, this research proposes a Virtual Reality augmented visual feedback system with haptic simulators to generate immersive, intuitive and effective feedback for human operators. The system can generate visual and haptic feedback as indications of flow conditions based on sensory data. A human subject experiment was performed to verify the effectiveness of this system. The result indicated that this system could significantly assist human operators in precepting the flow conditions in the ROV workspace in a more immersive and intuitive way.