Dynamic Stopping Research Articles

Comparing P300 flashing paradigms in online typing with language models.

The P300 Speller is a brain-computer interface system that allows victims of motor neuron diseases to regain the ability to communicate by typing characters into a computer by thought. Since the system has a relatively slow typing speed, different stimulus presentation paradigms have been proposed designed to allow users to input information faster by reducing the number of required stimuli or increase signal fidelity. This study compares the typing speeds of the Row-Column, Checkerboard, and Combinatorial Paradigms to examine how their performance compares in online and offline settings. When the different flashing patterns were tested in conjunction with other established optimization techniques such as language models and dynamic stopping, they did not make a significant impact on P300 speller performance. This result could indicate that further performance improvements on the system lie beyond optimizing flashing patterns.

A Bayesian dynamic stopping method for evoked response brain-computer interfacing.

As brain-computer interfacing (BCI) systems transition fromassistive technology to more diverse applications, their speed, reliability, and user experience become increasingly important. Dynamic stopping methods enhance BCI system speed by deciding at any moment whether to output a result or wait for more information. Such approach leverages trial variance, allowing good trials to be detected earlier, thereby speeding up the process without significantly compromising accuracy. Existing dynamic stopping algorithms typically optimize measures such as symbols per minute (SPM) and information transfer rate (ITR). However, these metrics may not accurately reflect system performance for specific applications or user types. Moreover, many methods depend on arbitrary thresholds or parameters that require extensive training data. We propose a model-based approach that takes advantage of the analytical knowledge that we have about the underlying classification model. By using a risk minimization approach, our model allows precise control over the types of errors and the balance between precision and speed. This adaptability makes it ideal for customizing BCI systems to meet the diverse needs of various applications. We validate our proposed method on a publicly available dataset, comparing it with established static and dynamic stopping methods. Our results demonstrate that our approach offers a broad range of accuracy-speed trade-offs and achieves higher precision than baseline stopping methods.

PSO-based lightweight neural architecture search for object detection

In recent years, Neural Architecture Search (NAS) has received widespread attention for its remarkable ability to design neural networks. However, existing NAS works have mainly focused on network search, with limited emphasis on downstream applications after discovering efficient neural networks. In this paper, we propose a lightweight search strategy based on the particle swarm optimization algorithm and apply the searched network as backbone for object detection tasks. Specifically, we design a lightweight search space based on Ghostconv modules and improved Mobileblocks, achieving comprehensive exploration within the search space using variable-length encoding strategy. During the search process, to balance network performance and resource consumption, we propose a multi-objective fitness function and incorporated the classification accuracy, parameter size, and FLOPs of candidate individuals into optimization. For particle performance evaluation, we propose a new strategy based on weight sharing and dynamic early stopping, significantly accelerating the search process. Finally, we fine-tune the globally optimal particle decoded as the backbone, adding Ghost PAN feature fusion modules and detection heads to build an object detection model, and we achieve a 17.01% mAP on the VisDrone2019 dataset. Experimental results demonstrate the competitiveness of our algorithm in terms of search time and the balance between accuracy and efficiency, and also confirm the effectiveness of object detection models designed through NAS methods.

Advancing energy efficiency of spiking neural network accelerator via dynamic predictive early stopping

Advancing energy efficiency of spiking neural network accelerator via dynamic predictive early stopping

Novel Ransomware Detection Exploiting Uncertainty and Calibration Quality Measures Using Deep Learning

Ransomware poses a significant threat by encrypting files or systems demanding a ransom be paid. Early detection is essential to mitigate its impact. This paper presents an Uncertainty-Aware Dynamic Early Stopping (UA-DES) technique for optimizing Deep Belief Networks (DBNs) in ransomware detection. UA-DES leverages Bayesian methods, dropout techniques, and an active learning framework to dynamically adjust the number of epochs during the training of the detection model, preventing overfitting while enhancing model accuracy and reliability. Our solution takes a set of Application Programming Interfaces (APIs), representing ransomware behavior as input we call “UA-DES-DBN”. The method incorporates uncertainty and calibration quality measures, optimizing the training process for better more accurate ransomware detection. Experiments demonstrate the effectiveness of UA-DES-DBN compared to more conventional models. The proposed model improved accuracy from 94% to 98% across various input sizes, surpassing other models. UA-DES-DBN also decreased the false positive rate from 0.18 to 0.10, making it more useful in real-world cybersecurity applications.

Optimizing machine learning yield functions using query-by-committee for support vector classification with a dynamic stopping criterion

In the field of materials engineering, the accurate prediction of material behavior under various loading conditions is crucial. Machine Learning (ML) methods have emerged as promising tools for generating constitutive models straight from data, capable of describing complex material behavior in a more flexible way than classical constitutive models. Yield functions, which serve as foundation of constitutive models for plasticity, can be properly described in a data-oriented manner using ML methods. However, the quality of these descriptions heavily relies on the availability of sufficient high-quality and representative training data that needs to be generated by fundamental numerical simulations, experiments, or a combination of both. The present paper addresses the issue of data selection, by introducing an active learning approach for Support Vector Classification (SVC) and its application in training an ML yield function with suitable data. In this regard, the Query-By-Committee (QBC) algorithm was employed, guiding the selection of new training data points in regions of the feature space where a committee of models shows significant disagreement. This approach resulted in a marked reduction in the variance of model predictions throughout the active learning process. It was also shown that the rate of decrease in the variance went along with an increase in the quality of the trained model, quantified by the Matthews Correlation Coefficient (MCC). This demonstrated the effectiveness of the approach and offered us the possibility to define a dynamic stopping criterion based on the variance in the committee results.

Enhancing SSVEP-BCI Performance Under Fatigue State Using Dynamic Stopping Strategy.

Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) have emerged as a prominent technology due to their high information transfer rate, rapid calibration time, and robust signal-to-noise ratio. However, a critical challenge for practical applications is performance degradation caused by user fatigue during prolonged use. This work proposes novel methods to address this challenge by dynamically adjusting data acquisition length and updating detection models based on a fatigue-aware stopping strategy. Two 16-target SSVEP-BCIs were employed, one using low-frequency and the other using high-frequency stimulation. A self-recorded fatigue dataset from 24 subjects was utilized for extensive evaluation. A simulated online experiment demonstrated that the proposed methods outperform the conventional fixed stopping strategy in terms of classification accuracy, information transfer rate, and selection time, irrespective of stimulation frequency. These findings suggest that the proposed approach can significantly improve SSVEP-BCI performance under fatigue conditions, leading to superior performance during extended use.

Improving dynamic collision frequencies: Impacts on dynamic structure factors and stopping powers in warm dense matter

Simulations and diagnostics of high-energy-density plasmas and warm dense matter rely on models of material response properties, both static and dynamic (frequency-dependent). Here, we systematically investigate variations in dynamic electron–ion collision frequencies ν(ω) in warm dense matter using data from a self-consistent-field average-atom model. We show that including the full quantum density of states, strong collisions, and inelastic collisions lead to significant changes in ν(ω). These changes result in red shifts and broadening of the plasmon peak in the dynamic structure factor, an effect observable in x-ray Thomson scattering spectra, and modify stopping powers around the Bragg peak. These changes improve the agreement of computationally efficient average-atom models with first-principles time-dependent density functional theory in warm dense aluminum, carbon, and deuterium.

Multiobjective Collaborative Optimization Method for the Urban Rail Multirouting Train Operation Plan

The Train Operation Plan (TOP) of urban rail transit (URT) is a comprehensive plan for the operation of trains, the use of facilities and equipment, and the organization of other operational tasks. The TOP should not only be formulated in terms of time-varying passenger flow periods, but it should also be arranged to consider the substitutability of trains between multiple routes combined with the passenger choice. Based on the principle of “operating by the flow” and the requirement for precise allocation of transport capacity for multiple routes, this article constructs a multiobjective nonlinear integer programming model by taking the minimized generalized travel cost of passengers, total running mileage of trains, fluctuation of trains for each route (as optimization targets), and the combination of requirements of both headways and fully loaded rates as constraints. A multiobjective genetic-based algorithm is designed to simultaneously optimize the TOP and the two-way train stopping time in each period. Finally, the proposed model and algorithm are validated with the real data from the Guangzhou Metro Line 2. The results show that the Pareto optimal TOP and dynamic train stopping time are significantly improved compared to the original values.

BCI-Utility Metric for Asynchronous P300 Brain-Computer Interface Systems.

The Brain-Computer Interface (BCI) was envisioned as an assistive technology option for people with severe movement impairments. The traditional synchronous event-related potential (ERP) BCI design uses a fixed communication speed and is vulnerable to variations in attention. Recent ERP BCI designs have added asynchronous features, including abstention and dynamic stopping, but it remains a open question of how to evaluate asynchronous BCI performance. In this work, we build on the BCI-Utility metric to create the first evaluation metric with special consideration of the asynchronous features of self-paced BCIs. This metric considers accuracy as all of the following three - probability of a correct selection when a selection was intended, probability of making a selection when a selection was intended, and probability of an abstention when an abstention was intended. Further, it considers the average time required for a selection when using dynamic stopping and the proportion of intended selections versus abstentions. We establish the validity of the derived metric via extensive simulations, and illustrate and discuss its practical usage on real-world BCI data. We describe the relative contribution of different inputs with plots of BCI-Utility curves under different parameter settings. Generally, the BCI-Utility metric increases as any of the accuracy values increase and decreases as the expected time for an intended selection increases. Furthermore, in many situations, we find shortening the expected time of an intended selection is the most effective way to improve the BCI-Utility, which necessitates the advancement of asynchronous BCI systems capable of accurate abstention and dynamic stopping.

Enhancing Detection of Control State for High-Speed Asynchronous SSVEP-BCIs Using Frequency-Specific Framework.

This study proposed a novel frequency-specific (FS) algorithm framework for enhancing control state detection using short data length toward high-performance asynchronous steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCI). The FS framework sequentially incorporated task-related component analysis (TRCA)-based SSVEP identification and a classifier bank containing multiple FS control state detection classifiers. For an input EEG epoch, the FS framework first identified its potential SSVEP frequency using the TRCA-based method and then recognized its control state using one of the classifiers trained on the features specifically related to the identified frequency. A frequency-unified (FU) framework that conducted control state detection using a unified classifier trained on features related to all candidate frequencies was proposed to compare with the FS framework. Offline evaluation using data lengths within 1 s found that the FS framework achieved excellent performance and significantly outperformed the FU framework. 14-target FS and FU asynchronous systems were separately constructed by incorporating a simple dynamic stopping strategy and validated using a cue-guided selection task in an online experiment. Using averaged data length of 591.63±5.65 ms, the online FS system significantly outperformed the FU system and achieved an information transfer rate, true positive rate, false positive rate, and balanced accuracy of 124.95±12.35 bits/min, 93.16±4.4%, 5.21±5.85%, and 92.89±4.02%, respectively. The FS system was also of higher reliability by accepting more correctly identified SSVEP trials and rejecting more wrongly identified ones. These results suggest that the FS framework has great potential to enhance the control state detection for high-speed asynchronous SSVEP-BCIs.

Learning from Manipulable Signals

We study a dynamic stopping game between a principal and an agent. The principal gradually learns about the agent's private type from a noisy performance measure that can be manipulated by the agent via a costly and hidden action. We fully characterize the unique Markov equilibrium of this game. We find that terminations/market crashes are often preceded by a spike in manipulation intensity and (expected) performance. Moreover, due to endogenous signal manipulation, too much transparency can inhibit learning and harm the principal. As the players get arbitrarily patient, the principal elicits no useful information from the observed signal. (JEL C73, D82, D83, G24, M13)

Instant classification for the spatially-coded BCI.

The spatially-coded SSVEP BCI exploits changes in the topography of the steady-state visual evoked response to visual flicker stimulation in the extrafoveal field of view. In contrast to frequency-coded SSVEP BCIs, the operator does not gaze into any flickering lights; therefore, this paradigm can reduce visual fatigue. Other advantages include high classification accuracies and a simplified stimulation setup. Previous studies of the paradigm used stimulation intervals of a fixed duration. For frequency-coded SSVEP BCIs, it has been shown that dynamically adjusting the trial duration can increase the system's information transfer rate (ITR). We therefore investigated whether a similar increase could be achieved for spatially-coded BCIs by applying dynamic stopping methods. To this end we introduced a new stopping criterion which combines the likelihood of the classification result and its stability across larger data windows. Whereas the BCI achieved an average ITR of 28.4±6.4 bits/min with fixed intervals, dynamic intervals increased the performance to 81.1±44.4 bits/min. Users were able to maintain performance up to 60 minutes of continuous operation. We suggest that the dynamic response time might have worked as a kind of temporal feedback which allowed operators to optimize their brain signals and compensate fatigue.

Early Stopping in Experimentation With Real-Time Functional Magnetic Resonance Imaging Using a Modified Sequential Probability Ratio Test.

Introduction: Functional magnetic resonance imaging (fMRI) often involves long scanning durations to ensure the associated brain activity can be detected. However, excessive experimentation can lead to many undesirable effects, such as from learning and/or fatigue effects, discomfort for the subject, excessive motion artifacts and loss of sustained attention on task. Overly long experimentation can thus have a detrimental effect on signal quality and accurate voxel activation detection. Here, we propose dynamic experimentation with real-time fMRI using a novel statistically driven approach that invokes early stopping when sufficient statistical evidence for assessing the task-related activation is observed.Methods: Voxel-level sequential probability ratio test (SPRT) statistics based on general linear models (GLMs) were implemented on fMRI scans of a mathematical 1-back task from 12 healthy teenage subjects and 11 teenage subjects born extremely preterm (EPT). This approach is based on likelihood ratios and allows for systematic early stopping based on target statistical error thresholds. We adopt a two-stage estimation approach that allows for accurate estimates of GLM parameters before stopping is considered. Early stopping performance is reported for different first stage lengths, and activation results are compared with full durations. Finally, group comparisons are conducted with both early stopped and full duration scan data. Numerical parallelization was employed to facilitate completion of computations involving a new scan within every repetition time (TR).Results: Use of SPRT demonstrates the feasibility and efficiency gains of automated early stopping, with comparable activation detection as with full protocols. Dynamic stopping of stimulus administration was achieved in around half of subjects, with typical time savings of up to 33% (4 min on a 12 min scan). A group analysis produced similar patterns of activity for control subjects between early stopping and full duration scans. The EPT group, individually, demonstrated more variability in location and extent of the activations compared to the normal term control group. This was apparent in the EPT group results, reflected by fewer and smaller clusters.Conclusion: A systematic statistical approach for early stopping with real-time fMRI experimentation has been implemented. This dynamic approach has promise for reducing subject burden and fatigue effects.

Automobile Replacement and Government Subsidy: An Analysis of the CARS Program

Subsidy programs are widely used by governments to stimulate economic activity in durable goods sectors. This paper investigates the impact of government subsidy on consumers' automobile replacement decisions in the context of the U.S. Car Allowance and Rebate System program in 2009. I develop and estimate a random-coefficient discrete-choice model that incorporates a dynamic optimal stopping problem. Counterfactual simulations show that 65% of the households would replace their vehicles even without the subsidy. The subsidy accelerates sales by only 4 months, and the reductions in gasoline consumption and carbon emission are also found to be limited. To highlight the importance of subsidy design, I find that limiting the subsidy to low-income consumers could generate 85% of the sales with half amount of total government spending. These results emphasize the importance of balancing the policy stimulus and government spending, and targeting the consumers more efficiently.

Early stopping in seamless phase I/II clinical trials.

In recent years, seamless phase I/II clinical trials have drawn much attention, as they consider both toxicity and efficacy endpoints in finding an optimal dose (OD). Engaging an appropriate number of patients in a trial is a challenging task. This paper attempts a dynamic stopping rule to save resources in phase I/II trials. That is, the stopping rule aims to save patients from unnecessary toxic or subtherapeutic doses. We allow a trial to stop early when widths of the confidence intervals for the dose-response parameters become narrower or when the sample size is equal to a predefined size, whichever comes first. The simulation study of dose-response scenarios in various settings demonstrates that the proposed stopping rule can engage an appropriate number of patients. Therefore, we suggest its use in clinical trials.

Operate P300 speller when performing other task

Objective. The P300 speller is a classic brain–computer interface (BCI) paradigm that has the potential to restore impaired motor control function. However, previous studies have confirmed that the letter recognition accuracy (LRA) of the P300 speller is a challenge when performing other tasks. Approach. To address this, we implemented a dynamic stopping strategy (DSS) to maintain the P300 speller LRA when performing multiple tasks simultaneously. Multiple tasks with dynamic workload levels were adopted to simulate the brain’s other thinking activities while operating P300 speller. A Bayes-based DSS offline model was built in single-task (only P300 speller task) and an online P300 speller system was established to test the DSS algorithm feasibility in dual-task. Main results. Online experimental results showed that the P300 speller with DSS could achieve a high LRA (96.9%) under dual-task, which was similar to single-task (98.7%, p = 0.126). Under dual-task, DSS dynamically adjusted the discriminant confidence according to the workload levels of the distraction tasks (correlation coefficient r = −0.68). Therefore, DSS can increase the repeated sequences to compensate for the reduction of P300 speller signal-to-noise ratio caused by parallel thinking activities. The average of repeated sequences increased significantly from 4.98 times under single-task to 6.22 times under dual-task (p < 0.005). These results indicated that the P300 speller feature is robust and the DSS model built in single-task maintained the applicability in various dual-tasks. Significance. Overall, this study provides a basis for the implementation of laboratory-developed BCI in real-world environments.

Pulse time reversal and stopping by a refractive index front

We discuss how dynamic light stopping and pulse time reversal can be implemented in dispersive waveguides via indirect photonic transitions induced by moving refractive index fronts. The previous concepts of light stopping/time reversal either require complex local variation of the device’s refractive index or rely on the strict phase matching condition, which imposes limitations on the amount of manipulated information. Until now, only single pulses or continuous waves were manipulated experimentally. Our scheme is not limited by a strict phase matching condition and does not require local index variations, thus it can manipulate broadband signals in a single step process. Here, we present several numerically integrated results for pulse time reversal and stopping/storage via indirect front-induced transitions. The presented results are experimentally feasible using existing photonic waveguide technologies.

Optimizing SSVEP-Based BCI System towards Practical High-Speed Spelling.

The brain–computer interface (BCI) spellers based on steady-state visual evoked potentials (SSVEPs) have recently been widely investigated for their high information transfer rates (ITRs). This paper aims to improve the practicability of the SSVEP-BCIs for high-speed spelling. The system acquired the electroencephalogram (EEG) data from a self-developed dedicated EEG device and the stimulation was arranged as a keyboard. The task-related component analysis (TRCA) spatial filter was modified (mTRCA) for target classification and showed significantly higher performance compared with the original TRCA in the offline analysis. In the online system, the dynamic stopping (DS) strategy based on Bayesian posterior probability was utilized to realize alterable stimulating time. In addition, the temporal filtering process and the programs were optimized to facilitate the online DS operation. Notably, the online ITR reached 330.4 ± 45.4 bits/min on average, which is significantly higher than that of fixed stopping (FS) strategy, and the peak value of 420.2 bits/min is the highest online spelling ITR with a SSVEP-BCI up to now. The proposed system with portable EEG acquisition, friendly interaction, and alterable time of command output provides more flexibility for SSVEP-based BCIs and is promising for practical high-speed spelling.

Doubling the Speed of N200 Speller via Dual-Directional Motion Encoding.

Motion-onset visual evoked potentials (mVEPs)-based spellers, also known as N200 spellers, have been successfully implemented, avoiding flashing stimuli that are common in visual brain-computer interface (BCI). However, their information transfer rates (ITRs), typically below 50 bits/min, are lower than other visual BCI spellers. In this study, we sought to improve the speed of N200 speller to a level above the well-known P300 spellers. Based on our finding of the spatio-temporal asymmetry of N200 response elicited by leftward and rightward visual motion, a novel dual-directional N200 speller was implemented. By presenting visual stimuli moving in two different directions simultaneously, the new paradigm reduced the stimuli presentation time by half, while ensuring separable N200 features between two visual motion directions. Furthermore, a probability-based dynamic stopping algorithm was also proposed to shorten the decision time for each output further. Both offline and online tests were conducted to evaluate the performance in ten participants. Offline results revealed contralateral dominant temporal and spatial patterns in N200 responses when subjects attended to stimuli moving leftward or rightward. In online experiments, the dual-directional paradigm achieved an average ITR of 79.8 bits/min, with the highest ITR of 124.8 bits/min. Compared with the traditional uni-directional N200 speller, the median gain on the ITR was 202%. The proposed dual-directional paradigm managed to double the speed of the N200 speller. Together with its non-flashing characteristics, this dual-directional N200 speller is promising to be a competent candidate for fast and reliable BCI applications.