Adaptive Selection Scheme Research Articles

A compass-based hyper-heuristic for multi-objective optimization problems

Multi-objective selection hyper-heuristics have attracted more attention of researchers because of their cross-domain ability. However, for multi-objective optimization problems (MOPs), obtaining a manageable number of solutions that are well distributed and converged in the objective space is still a challenge, especially when solving high-dimensional MOPs. In order to solve this problem, this paper proposes a compass-based hyper-heuristics(COHH), which is a general iterative framework that learns and selects from a set of meta-heuristics or components (named low-level heuristics, LLHs). The selected LLH is applied to solve the given MOP at the current iteration. In order to learn the potential of LLHs, the impact of the diversity of the current solution set on the final performance is studied. Then a new compass-based indicator is defined to evaluate the current solution sets. The learning strategy with new indicator can bias to diversity by adjusting the angle of a reference vector. After learning, the adaptive two-stage selection strategy triggered by the quality of the current solution set is used to choose LLH. Experiments are conducted on DTLZ, MaOP, WFG, and MaF test suites, as well as several real-world constrained test problems. Experimental results show that COHH is competitive in performance and cross-domain capability when compared with popular meta-heuristics and hyper-heuristics.

Optimal clustering from noisy binary feedback

AbstractWe study the problem of clustering a set of items from binary user feedback. Such a problem arises in crowdsourcing platforms solving large-scale labeling tasks with minimal effort put on the users. For example, in some of the recent reCAPTCHA systems, users clicks (binary answers) can be used to efficiently label images. In our inference problem, items are grouped into initially unknown non-overlapping clusters. To recover these clusters, the learner sequentially presents to users a finite list of items together with a question with a binary answer selected from a fixed finite set. For each of these items, the user provides a noisy answer whose expectation is determined by the item cluster and the question and by an item-specific parameter characterizing the hardness of classifying the item. The objective is to devise an algorithm with a minimal cluster recovery error rate. We derive problem-specific information-theoretical lower bounds on the error rate satisfied by any algorithm, for both uniform and adaptive (list, question) selection strategies. For uniform selection, we present a simple algorithm built upon the K-means algorithm and whose performance almost matches the fundamental limits. For adaptive selection, we develop an adaptive algorithm that is inspired by the derivation of the information-theoretical error lower bounds, and in turn allocates the budget in an efficient way. The algorithm learns to select items hard to cluster and relevant questions more often. We compare the performance of our algorithms with or without the adaptive selection strategy numerically and illustrate the gain achieved by being adaptive.

Scattering-Point-Guided Oriented RepPoints for Ship Detection

Ship detection finds extensive applications in fisheries management, maritime rescue, and surveillance. However, detecting nearshore targets in SAR images is challenging due to land scattering interference and non-axisymmetric ship shapes. Existing SAR ship detection models struggle to adapt to oriented ship detection in complex nearshore environments. To address this, we propose an oriented-reppoints target detection scheme guided by scattering points in SAR images. Our method deeply integrates SAR image target scattering characteristics and designs an adaptive sample selection scheme guided by target scattering points. This incorporates scattering position features into the sample quality measurement scheme, providing the network with a higher-quality set of proposed reppoints. We also introduce a novel supervised guidance paradigm that uses target scattering points to guide the initialization of reppoints, mitigating the influence of land scattering interference on the initial reppoints quality. This achieves adaptive feature learning, enhancing the quality of the initial reppoints set and the performance of object detection. Our method has been extensively tested on the SSDD and HRSID datasets, where we achieved mAP scores of 89.8% and 80.8%, respectively. These scores represent significant improvements over the baseline methods, demonstrating the effectiveness and robustness of our approach. Additionally, our method exhibits strong anti-interference capabilities in nearshore detection and has achieved state-of-the-art performance.

Image restoration based on transformed total variation and deep image prior

Most supervised learning methods require observation data and ground truth pairs as data sets to train the network. However, it is difficult and time-consuming to obtain a large number of high quality data sets, because ground truth is not available in some practical settings, such as medical imaging, dynamic scenes. Deep image prior (DIP) only uses one degraded image for image recovery tasks, which gets rid of the limitation of constructing a large number of training sets but requires an early stop mechanism. In order to further improve the image restoration ability of the DIP model, we combine it with the transformed total variation (TTV) regularization, which is a generalization of the classical total variation (TV) regularization and can achieve better performance for image restoration problems. The proposed method not only uses the deep neural network to capture image prior, but also exploits inherent sparse prior in image gradient domain. In addition, we provide an adaptive weight selection strategy for TTV regularization. The ADMM scheme is employed to solve the proposed models. Numerical results of image restoration illustrate that the proposed methods perform better than the compared methods.

Investigation of the Variability of Dry / Wet Meteorological conditions using the SPEI / SPI Index in the Selo Watershed, Period 1981-2020.

The significant impact of climate change is high variability in rainfall and an increase in extreme rainfall events. The impacts resulted in severe dry/wet conditions. Severe dry conditions can lead to crop failure due to lack of water availability, while severe wet conditions can indicate flooding. This paper aims to investigate the variability of dry/wet meteorological conditions in the Selo watershed. The investigation was carried out using the SPEI and SPI. The results show that wet/dry conditions do not have a clear pattern variability in the long term. Seasonal wet/dry conditions show a changing trend towards wetter conditions in all seasons. Spatially, the changes occur throughout the study area, with the most significant changes occurring in the upstream and downstream parts of the watershed. In the last two decades, wet conditions are more dominant to replace dry conditions that were more prominent in the previous two decades. Other analysis results conclude that there is no influence of air temperature in creating wet/dry conditions in the Selo watershed area. So that the characteristics of wet/dry conditions are very useful for considering the selection of agricultural adaptation strategies in dealing with variability and climate change in the Selo watershed area.

Adapt only once: Fast unsupervised person re-identification via relevance-aware guidance

Unsupervised domain adaptive person re-identification (UDA person reID) defines a task where labels in target domains are totally unknown while source domains are fully labeled. Assigning reliable labels quickly is a critical issue for UDA person reID that could be applied in the real-world scenarios. Recent studies focus on obtaining pseudo labels by clustering algorithms and then training the reID model with these labels. However, the main limitation of these methods is the high time complexity, which is caused by the calculation of all pair-wise similarities and multiple iterations in the clustering algorithm to obtain reliable results. When the data is very large or the feature dimensions are very high, the memory and time cost requirements of the clustering algorithm can increase rapidly.In this paper, we provide a fast unsupervised domain adaptive person reID framework (FUReID), which calculates the relevance between unlabeled samples only once to adapt to the new scenarios without any iterations in the stage of label generation. Especially, instead of pursuing accurate labels, FUReID considers constructing a lightweight paradigm to generate coarse labels and then refine these labels during the training stage. Therefore, FUReID designs a prototype-guided labeling method that only relies on calculating the relevance between the prototype vectors and the samples, and assigning coarse labels with noise. Then, to alleviate the issue of noise, FUReID designs a label-flexible training network with an adaptive selection strategy to refine those coarse labels progressively. For several widely-used person reID datasets, our method achieves 81.7%, 26.2%, and 87.7% in mAP on Market1501, MSMT17 and PersonX, respectively. Code is available at https://github.com/AILab90/FUReID.

An artificial bee colony algorithm with an adaptive search strategy selection mechanism and its application on workload prediction

As one category of evolutionary algorithms, artificial bee colony (ABC) algorithms have many advantages such as few control parameters, simple structures and competitive performance. However, most ABC algorithms suffer from slow convergence and poor performance on optimization problems with strong correlations. To overcome these limitations, we propose a novel ABC with an Adaptive Search Strategy Selection (AS3) mechanism, called ABC-AS3. The basic idea of ABC-AS3 is to adaptively adjust search strategies between a one-dimensional search strategy and an all-dimensional search strategy based on probabilities that they are selected respectively within the most recent generations. We prove the rotation-variant and rotation-invariant properties of the one-dimensional and all-dimensional search strategies applied in the ABC-AS3 algorithm, respectively. The one-dimensional search strategy is used to maintain superiority on optimization with weak correlations. The all-dimensional one aims to accelerate the convergence speed and improve the performance on optimization with strong correlations. Simulation results show that ABC-AS3 outperforms state-of-the-art ABC variants and evolutionary algorithms in terms of solution quality, convergence speed, and numerical stability. In addition, we apply ABC-AS3 to optimize parameters of a multi-layer perceptron for workload prediction in data centers, which achieves better prediction accuracy and stability than baseline methods.

Adaptive Point-Line Fusion: A Targetless LiDAR-Camera Calibration Method with Scheme Selection for Autonomous Driving.

Accurate calibration between LiDAR and camera sensors is crucial for autonomous driving systems to perceive and understand the environment effectively. Typically, LiDAR-camera extrinsic calibration requires feature alignment and overlapping fields of view. Aligning features from different modalities can be challenging due to noise influence. Therefore, this paper proposes a targetless extrinsic calibration method for monocular cameras and LiDAR sensors that have a non-overlapping field of view. The proposed solution uses pose transformation to establish data association across different modalities. This conversion turns the calibration problem into an optimization problem within a visual SLAM system without requiring overlapping views. To improve performance, line features serve as constraints in visual SLAM. Accurate positions of line segments are obtained by utilizing an extended photometric error optimization method. Moreover, a strategy is proposed for selecting appropriate calibration methods from among several alternative optimization schemes. This adaptive calibration method selection strategy ensures robust calibration performance in urban autonomous driving scenarios with varying lighting and environmental textures while avoiding failures and excessive bias that may result from relying on a single approach.

A new MPPT controller based on a modified multiswarm PSO algorithm using an adaptive factor selection strategy for partially shaded PV systems

Maximum power point tracking (MPPT) controller is the main element in photovoltaic (PV) systems, which is used to ensure maximum power extraction under different meteorological conditions. A MPPT controller can guarantee good performance criteria even in the presence of climatic changes. To achieve this goal, several techniques have been proposed in the literature to improve robustness of the PV system control, such as artificial intelligence and multiswarm particle swarm optimization (MSPSO) algorithm. Previous research on classical MSPSO has shown that the algorithm search behavior cannot find the optimal solution for certain problems. In this context, we investigate the design of a new MPPT controller based on a modified version of heterogeneous multiswarm particle swarm optimization algorithm using an adaptive factor selection strategy (FMSPSO) for PV systems. The proposed FMSPSO can improve the tracking capability with high accuracy, less oscillations, and high robustness. To validate the proposed solution, a simulation and experimental benchmarking of a PV system are presented and analyzed. The obtained results show the effectiveness of the proposed solution compared with the classical MSPSO, fuzzy logic, and perturb and observe (P&O) control presented in other recent works.

Adaptive Interference Avoidance and Mode Selection Scheme for D2D-Enabled Small Cells in 5G-IIoT Networks

Adaptive Interference Avoidance and Mode Selection Scheme for D2D-Enabled Small Cells in 5G-IIoT Networks

Safety, feasibility, tolerability, and clinical effects of repeated psilocybin dosing combined with non-directive support in the treatment of obsessive-compulsive disorder: protocol for a randomized, waitlist-controlled trial with blinded ratings.

To date, few randomized controlled trials of psilocybin with non-directive support exist for obsessive-compulsive disorder (OCD). Results and participant feedback from an interim analysis of an ongoing single-dose trial (NCT03356483) converged on the possibility of administering a higher fixed dose and/or more doses of psilocybin in future trials for presumably greater benefits. This trial aims to evaluate the safety, feasibility, tolerability, and clinical effects of two doses of psilocybin paired with non-directive support in the treatment of OCD. This trial also seeks to examine whether two doses of psilocybin lead to greater OCD symptom reduction than a single dose, and to elucidate psychological mechanisms underlying the effects of psilocybin on OCD. A randomized (1:1), waitlist-controlled design with blinded ratings will be used to examine the effects of two doses of oral psilocybin paired with non-directive support vs. waitlist control on OCD symptoms. An adaptive dose selection strategy will be implemented (i.e., first dose: 25 mg; second dose: 25 or 30 mg). This single-site trial will enroll 30 adult participants with treatment-refractory OCD. Aside from safety, feasibility, and tolerability metrics, primary outcomes include OCD symptoms assessed on the Yale-Brown Obsessive-Compulsive Scale - Second Edition (Y-BOCS-II). A blinded independent rater will assess primary outcomes at baseline and the primary endpoint at the end of the second dosing week. Participants will be followed up to 12 months post-second dosing. Participants randomized to waitlist will be rescreened after 7 weeks post-randomization, and begin their delayed treatment phase thereafter if still eligible. Written informed consent will be obtained from participants. The institutional review board has approved this trial (protocol v. 1.7; HIC #2000032623). This study seeks to advance our ability to treat refractory OCD, and catalyze future research seeking to optimize the process of psilocybin treatment for OCD through understanding relevant psychological mechanisms.Clinical trial registration: ClinicalTrials.gov, identifier NCT05370911.

Distilling object detectors with efficient logit mimicking and mask-guided feature imitation

Knowledge distillation (KD) is a promising approach to learning compact models for object detection with information inherited from intricate teacher networks. In this paper, we raise some shortcomings of existing KD methods for object detectors, e.g., ignoring knowledge selection, coarse feature imitation mask, etc. To address these issues, a novel KD framework has been presented to train efficient object detectors via Logit Mimicking and Feature Imitation (LMFI). First, a novel logit mimicking method is put forward to distill classification and localization heads. On the one hand, it first proposes to mimic the classification logits of one-category object detectors. On the other hand, the localization knowledge from teacher predictions and ground truths are exploited, which dynamically guides the learning of student’s regression outputs by stages. Second, an adaptive positive teacher selection (APTS) strategy is designed to obtain high-quality teacher samples during distillation, which reduces the transmission of inferior knowledge. Moreover, a soft metric and a fine-grained mask are heuristically introduced to reconcile the discrepancy between teacher and student features in a position-wise manner. Extensive experiments show that LMFI outperforms the state-of-the-art KD frameworks for object detection. It can significantly boost the performance of various detectors on different benchmarks, e.g., 2.83% and 2.58% MR−2 reduction of Cascade R-CNN on the S and All sets of CityPersons, and an improvement of ResNet-50 based TOOD from 40.3% to 42.9% mAP on the COCO benchmark.

Application of IoT technology in cyber security prevention system

Abstract In the process of gradually expanding the scale of computer networks and the design of network systems becoming more and more complex, people pay more and more attention to the construction of network security protection systems. Starting from the blockchain encryption technology, the article establishes the authentication and access management key based on the elliptic curve encryption algorithm and combines the maximum entropy model with the hidden Markov model to construct the MEMM for intrusion detection of network security. Based on the effective signal-to-noise ratio model of the network channel, an adaptive channel selection strategy based on the UCB algorithm is proposed. The IoT security prevention system is built based on IoT technology, and each functional module of the system is designed. The system’s authentication security, network intrusion detection, adaptive channel selection, and concurrency performance were tested after the design was completed. The encryption operation time of the ECC algorithm was improved by 41.53% compared to the RSA algorithm, the average time of the MEMM network intrusion detection was 41.54ms, and the false alarm rate of the intrusion detection was kept below 16.5%. The average packet collection rate of the nodes in the adaptive channel selection algorithm is 90.98%. The maximum system throughput is up to 62.19MB, and the extreme difference in data volume between different nodes is only 38 entries. Constructing a network security prevention system based on IoT technology and combining multiple encryption techniques can ensure the secure transmission of network data.

Machine Learning‐Based Distraction‐Free Method for Measuring the Optical Displacement of Long‐Span Bridge Structures

A photogrammetric displacement measurement method based on machine learning was proposed to improve the robustness to environmental disturbances. (1) To reduce the target positioning error caused by environmental vibration especially atmospheric turbulence, a machine learning‐based weighted location algorithm combined with an adaptive window selection strategy was developed. In an outdoor displacement table experiment, the proposed method’s root mean squared error (RMSE) is 0.04 mm when the distance is 50 m, showing better accuracy and stability. (2) To complement/correct the missing or anomalous data caused by adverse external conditions, such as severe occlusion or camera shaking, a fast data self‐diagnosis using a correlation vector machine was performed to make full use of the full‐field measurement results obtained from the images. The applicability of the proposed method in extreme cases was demonstrated in designed experiments and an actual displacement measurement task of a long‐span bridge subjected to vortex‐induced vibration.

Adaptive multi-source domain collaborative fine-tuning for transfer learning.

Fine-tuning is an important technique in transfer learning that has achieved significant success in tasks that lack training data. However, as it is difficult to extract effective features for single-source domain fine-tuning when the data distribution difference between the source and the target domain is large, we propose a transfer learning framework based on multi-source domain called adaptive multi-source domain collaborative fine-tuning (AMCF) to address this issue. AMCF utilizes multiple source domain models for collaborative fine-tuning, thereby improving the feature extraction capability of model in the target task. Specifically, AMCF employs an adaptive multi-source domain layer selection strategy to customize appropriate layer fine-tuning schemes for the target task among multiple source domain models, aiming to extract more efficient features. Furthermore, a novel multi-source domain collaborative loss function is designed to facilitate the precise extraction of target data features by each source domain model. Simultaneously, it works towards minimizing the output difference among various source domain models, thereby enhancing the adaptability of the source domain model to the target data. In order to validate the effectiveness of AMCF, it is applied to seven public visual classification datasets commonly used in transfer learning, and compared with the most widely used single-source domain fine-tuning methods. Experimental results demonstrate that, in comparison with the existing fine-tuning methods, our method not only enhances the accuracy of feature extraction in the model but also provides precise layer fine-tuning schemes for the target task, thereby significantly improving the fine-tuning performance.

Unified Flowing Normality Learning for Rotating Machinery Anomaly Detection in Continuous Time-Varying Conditions.

Intelligent anomaly detection (AD) methods have achieved much successes in machinery condition monitoring. However, the underlying independent and identically distributed assumption restricts their application scopes to steady operating conditions. False and missing alarms would occur when machines operate under time-varying circumstances. In this work, a more challenging time-varying setting is studied, where the working conditions are continuously changing, such that few or no samples are available for model training at one single condition. To tackle this issue, we propose a unified flowing normality learning (UFNL) framework, which aims to capture the flowing normal conditional distribution of time-varying samples and assigns dynamic decision boundary for AD. Specifically, a manifold-based probability density estimation is utilized to guide the adversarial learning process of generative adversarial networks, where adjacent samples are aggregated to approximate the conditional distribution by a conditional generator. Then, a latent normality inversion is proposed to extract the manifold structure from the pretrained generator and to map it into the latent space via a conditional encoder. The reconstruction errors from the encoder and generator can reveal the deviation of signals to the flowing normality. Finally, a condition-aware adaptive threshold selection strategy is proposed, where different thresholds are adaptively assigned for different conditions. Experiments are carried out under two typical continuous time-varying scenarios. The results demonstrate that the proposed framework can realize accurate fault detection at any operating condition within continuously changing environments.

Functional responses in American marten habitat selection indicate cumulative effects of progressive habitat change

AbstractShifts in habitat selection may serve as behavioral indicators of changing habitat quality and can provide an insight into the effects of chronic disturbance on wildlife populations before rates of occurrence and other metrics related to population viability are affected. Long‐term studies can provide unique opportunities to understand animals' behavioral responses to the cumulative effects of habitat disturbance that may be missed by short‐term studies or space‐for‐time substitution. Using the American marten (Martes americana) as a focal species, we evaluated the effects of progressive habitat disturbance on an animal population by identifying changes in habitat availability and selection over time. Marten are strongly associated with mature forest conditions and are thus sensitive to forest harvesting, which is a chronic and prevalent form of disturbance across the species' range. We developed resource selection functions that characterized both temporally consistent patterns and functional responses in the relative selection of habitats by marten at the patch scale. We used a combination of marten location data collected during three periods over 30 years of extensive forest change in northern Maine and a habitat classification scheme informed by species' behavior and designed to capture habitat changes over time. Marten increased their selection of tall (>12 m mean tree height) mature forest, including uncut forest and regenerated clearcuts, in response to decreased availability. Marten also increased avoidance of early‐successional forest (<9 m mean tree height) in response to increased availability. These functional responses in habitat selection appear to enable marten to maintain access to required resources in mature forest and limit exposure to risky or resource‐poor habitat within home ranges. However, a declining trend in the occurrence of resident adult females may indicate a limit to the functional response strategies for individuals unable to incur the energetic or prey‐related costs that may be associated with strong functional responses. Our results suggest that, despite adaptive habitat selection strategies, there are minimum thresholds of mature forest availability below which resident adults, particularly females, will struggle to persist. We recommend forest management strategies targeted at minimizing the strength of functional responses displayed by marten within occupied landscapes.

Reinforcement Learning-Based Multi-Objective of Two-Stage Blocking Hybrid Flow Shop Scheduling Problem

Consideration of upstream congestion caused by busy downstream machinery, as well as transportation time between different production stages, is critical for improving production efficiency and reducing energy consumption in process industries. A two-stage hybrid flow shop scheduling problem is studied with the objective of the makespan and the total energy consumption while taking into consideration blocking and transportation restrictions. An adaptive objective selection-based Q-learning algorithm is designed to solve the problem. Nine state characteristics are extracted from real-time information about jobs, machines, and waiting processing queues. As scheduling actions, eight heuristic rules are used, including SPT, FCFS, Johnson, and others. To address the multi-objective optimization problem, an adaptive objective selection strategy based on t-tests is designed for making action decisions. This strategy can determine the optimization objective based on the confidence of the objective function under the current job and machine state, achieving coordinated optimization for multiple objectives. The experimental results indicate that the proposed algorithm, in comparison to Q-learning and the non-dominated sorting genetic algorithm, has shown an average improvement of 4.19% and 22.7% in the makespan, as well as 5.03% and 9.8% in the total energy consumption, respectively. The generated scheduling solutions provide theoretical guidance for production scheduling in process industries such as steel manufacturing. This contributes to helping enterprises reduce blocking and transportation energy consumption between upstream and downstream.

Feature-based search space characterisation for data-driven adaptive operator selection

Combinatorial optimisation problems are known as unpredictable and challenging due to their nature and complexity. One way to reduce the unpredictability of such problems is to identify features and the characteristics that can be utilised to guide the search using domain-knowledge and act accordingly. Many problem solving algorithms use multiple complementary operators in patterns to handle such unpredictable cases. A well-characterised search space may help to evaluate the problem states better and select/apply a neighbourhood operator to generate more productive new problem states that allow for a smoother path to the final/optimum solutions. This applies to the algorithms that use multiple operators to solve problems. However, the remaining challenge is determining how to select an operator in an optimal way from the set of operators while taking the search space conditions into consideration. Recent research shows the success of adaptive operator selection to address this problem. However, efficiency and scalability issues persist in this regard. In addition, selecting the most representative features remains crucial in addressing problem complexity and inducing commonality for transferring experience across domains. This paper investigates if a problem can be represented by a number of features identified by landscape analysis, and whether an adaptive operator selection scheme can be constructed using Machine Learning (ML) techniques to address the efficiency and scalability problem. The proposed method determines the optimal categorisation by analysing the predictivity of a set of features using the most well-known supervised ML techniques. The identified set of features is then used to construct an adaptive operator selection scheme. The findings of the experiments demonstrate that supervised ML algorithms are highly effective when building adaptable operator selectors.

Real-time implementation of a novel MPPT control based on the improved PSO algorithm using an adaptive factor selection strategy for photovoltaic systems

Particle Swarm Optimization (PSO) is considered as one of Maximum Power Point Tracking (MPPT) controller algorithm developed for PhotoVoltaic system (PV) to guarantee a maximum power extraction under different climatic conditions of temperature and irradiation. Classical MPPT algorithms like Perturbe and Observe (P&O), PSO, Adaptive Neuro-Fuzzy Inference System (ANFIS) are an effective method for tracking the maximum power point (MPP) for the PV systems, but the problems with these approaches that they are less stable, high oscillation around steady state and slower convergence to the MPP. Based on recent research, the purpose of this paper is to introduces a novel MPPT controller based on a modified version of heterogeneous multi swarm PSO algorithm using an adaptive factor selection strategy (FMSPSO), to overcome the previous shortcomings and compared with conventional PSO, ANFIS and classical P&O controllers. Simulation and experimental results revealed that the new FMSPSO algorithm can overcome the previous shortcomings providing the superior performance to track the MPP efficiently with a shorter convergence time and small oscillations compared to other algorithms. The experimental confirmation of the FMSPSO algorithm has been carried out using NI-myRIO-1900 card and shows that with the proposed MPPT approach efficiency can reach a value greater than 99% even in climatic variation of irradiation and temperature.