Look-ahead Mechanism Research Articles

A lightweight approach to real-time speaker diarization: from audio toward audio-visual data streams

This manuscript deals with the task of real-time speaker diarization (SD) for stream-wise data processing. Therefore, in contrast to most of the existing papers, it considers not only the accuracy but also the computational demands of individual investigated methods. We first propose a new lightweight scheme allowing us to perform speaker diarization of streamed audio data. Our approach utilizes a modified residual network with squeeze-and-excitation blocks (SE-ResNet-34) to extract speaker embeddings in an optimized way using cached buffers. These embeddings are subsequently used for voice activity detection (VAD) and block-online k-means clustering with a look-ahead mechanism. The described scheme yields results similar to the reference offline system while operating solely on a CPU with a low real-time factor (RTF) below 0.1 and a constant latency of around 5.5 s. In the next part of the work, our research moves toward much more demanding and complex real-time processing of audio-visual data streams. For this purpose, we extend the above-mentioned scheme for audio data processing by adding an audio-video module. This module utilizes SyncNet combined with visual embeddings for identity tracking. Our resulting multi-modal SD framework then combines the outputs from audio and audio-video modules by using a new overlap-based fusion strategy. It yields diarization error rates that are competitive with the existing state-of-the-art offline audio-visual methods while allowing us to process various audio-video streams, e.g., from Internet or TV broadcasts, in real-time using GPU and with the same latency as for audio data processing.

Realtime polymorphic malicious behavior detection in blockchain-based smart contracts

Abstract This paper proposes an innovative approach to achieving real-time polymorphic behavior detection, and its direct application to blockchain-focused smart-contracts. We devise a method based on a non-deterministic finite state machine to perform approximate pattern-matching, using a look-ahead mechanism implemented through a concept similar to that of a sliding window, and using threshold-based similarity checking at every state in the automaton. We introduce and formalize our approach, discuss the challenges we faced and then test it in a real-world environment. The experimental results obtained showed a significant speed-up of our approach, as compared to the classic similarity measures used commonly in such scenarios.

Enhancing Autonomous Vehicle Navigation with a Clothoid-Based Lateral Controller

This study introduces an advanced lateral control strategy for autonomous vehicles using a clothoid-based approach integrated with an adaptive lookahead mechanism. The primary focus is on enhancing lateral stability and path-tracking accuracy through the application of Euler spirals for smooth curvature transitions, thereby reducing passenger discomfort and the risk of vehicle rollover. An innovative aspect of our work is the adaptive adjustment of lookahead distance based on real-time vehicle dynamics and road geometry, which ensures optimal path following under varying conditions. A quasi-feedback control algorithm constructs optimal clothoids at each time step, generating the appropriate steering input. A lead filter compensates for the vehicle’s lateral dynamics lag, improving control responsiveness and stability. The effectiveness of the proposed controller is validated through a comprehensive co-simulation using TruckSim® and Simulink®, demonstrating significant improvements in lateral control performance across diverse driving scenarios. Future directions include scaling the controller for higher-speed applications and further optimization to minimize off-track errors, particularly for articulated vehicles.

Ant Colony Optimization With Look-Ahead Mechanism for Dynamic Traffic Signal Control of IoV Systems

In the era of Internet of Vehicle (IoV), traffic data are obtained by wireless sensor networks (WSNs), which provides convenience for real-time traffic signal control and is expected to further improve the efficiency of the whole system. Traffic signal controllers designed for IoV systems usually fully depend on data transmitted from the IoV system and are limited to the penetration rate of connected vehicles. This work proposes a dynamic ant colony optimization algorithm with the look-ahead mechanism (DALM) with eight no-conflict phases. The needed traffic information in the DALM algorithm is small and can be obtained independently by the system, which makes this method does not fully rely on the vehicle trajectories data and can better protect customer privacy and deal with cyber attack. With the look-ahead mechanism, the forthcoming traffic flow together with the current waiting queue are involved to predict the length of waiting vehicles and the optimal phase sequence is dynamically selected according to shortest green-light duration of candidate phase sequences obtained form the collected real-time traffic data. Experiments conducted under five scenarios with different kinds of traffic flow on SUMO platform show that compared with other four methods, both the total waiting time and the total carbon emissions of vehicles at the intersection are greatly reduced by using the DALM algorithm, especially in high traffic flow situations. The proposed algorithm can greatly improve the efficiency of the intersection in high traffic flow situation and is predictable to benefit the real-time traffic signal control in IoV system.

Identifying Anomalies while Preserving Privacy.

Identifying anomalies in data is vital in many domains, including medicine, finance, and national security. However, privacy concerns pose a significant roadblock to carrying out such an analysis. Since existing privacy definitions do not allow good accuracy when doing outlier analysis, the notion of sensitive privacy has been recently proposed to deal with this problem. Sensitive privacy makes it possible to analyze data for anomalies with practically meaningful accuracy while providing a strong guarantee similar to differential privacy, which is the prevalent privacy standard today. In this work, we relate sensitive privacy to other important notions of data privacy so that one can port the technical developments and private mechanism constructions from these related concepts to sensitive privacy. Sensitive privacy critically depends on the underlying anomaly model. We develop a novel n-step lookahead mechanism to efficiently answer arbitrary outlier queries, which provably guarantees sensitive privacy if we restrict our attention to common a class of anomaly models. We also provide general constructions to give sensitively private mechanisms for identifying anomalies and show the conditions under which the constructions would be optimal.

Surrogate-assisted hyper-parameter search for portfolio optimisation: multi-period considerations

AbstractPortfolio management is a multi-period multi-objective optimisation problem subject to various constraints. However, portfolio management is treated as a single-period problem partly due to the computationally burdensome hyper-parameter search procedure needed to construct a multi-period Pareto frontier. This study presents the Pareto driven surrogate (ParDen-Sur) modelling framework to efficiently perform the required hyper-parameter search. ParDen-Sur extends previous surrogate frameworks by including a reservoir sampling-based look-ahead mechanism for offspring generation in evolutionary algorithms (EAs) alongside the traditional acceptance sampling scheme. We evaluate this framework against, and in conjunction with, several seminal multi-objective (MO) EAs on two datasets for both the single- and multi-period use cases. When considering hypervolume ParDen-Sur improves marginally (0.8%) over the state-of-the-art (SOTA)-NSGA-II. However, for generational distance plus and inverted generational distance plus, these improvements over the SOTA are 19.4% and 66.5%, respectively. When considering the average number of evaluations and generations to reach a 99% success rate, ParDen-Sur is shown to be 1.84× and 2.02× more effective than the SOTA. This improvement is statistically significant for the Pareto frontiers, across multiple EAs, for both datasets and use cases.

A Dynamic Holding Approach to Stabilizing a Bus Line Based on the Q-Learning Algorithm with Multistage Look-Ahead

The unreliable service and the unstable operation of a high-frequency bus line are shown as bus bunching and the uneven distribution of headways along the bus line. Although many control strategies, such as the static and dynamic holding strategies, have been proposed to solve the above problems, many of them take on some oversimplified assumptions about the real bus line operation. So it is hard for them to continuously adapt to the evolving complex system. In view of this dynamic setting, we present an adaptive holding method that combines the classic approximate dynamic programming (ADP) with the multistage look-ahead mechanism. The holding time, the only control means used in this study, will be determined by estimating its impact on the operation stability of the bus line system in the remaining observation period. The multistage look-ahead mechanism introduced into the classic Q-learning algorithm of the ADP model makes it easy that the algorithm gets through its earlier unstable phase more quickly and easily. During the implementation of the new holding approach, the past experiences of holding operations can be cumulated effectively into an artificial neural network used to approximate the unavailable Q-factor. The use of a detailed simulation system in the new approach makes it possible to take into account most of the possible causes of instability. The numerical experiments show that the new holding approach can stabilize the system by producing evenly distributed headway and removing bus bunching thoroughly. Compared with the terminal station holding strategies, the new method brings a more reliable bus line with shorter waiting times for passengers.

A lookahead matheuristic for the unweighed variable-sized two-dimensional bin packing problem

The unweighed oriented variable-sized two-dimensional guillotine bin packing problem consists in packing without overlap small rectangular items into large non-identical rectangular bins, with the items obtained via guillotine cuts. It minimizes the waste of the used bins. It is herein approximately solved using a hybrid matheuristic, which applies a sequence of low-level mixed-integer programs that reserve space for unpacked items and that are guided by feasibility constraints and by upper bounds on the objective function. The embedded constraints constitute a lookahead mechanism that prohibits the investigation of infeasible directions and constrains the search to improving ones. The matheuristic further employs high-level diversification and intensification mechanisms. The diversification incorporates a sequential value correction algorithm that tags a pseudo-price to each item to govern the fitness functions of mixed integer programs and subsequently their solution construction process. The intensification is a local search that investigates the neighbourhood of promising solutions. The extensive computational experiments provide evidence of the good performance of the proposed matheuristic. For the variable-sized bin packing benchmark instances, the matheuristic matches and improves 90.8% of the upper bounds. For the single bin-size bin packing benchmark instances, the matheuristic further proves the optimality of 82.6% upper bounds while it matches 14.4% and improves 2.6% existing bounds.

High-dimensional microarray dataset classification using an improved adam optimizer (iAdam)

Classifying data samples into their respective categories is a challenging task, especially when the dataset has more features and only a few samples. A robust model is essential for the accurate classification of data samples. The logistic sigmoid model is one of the simplest model for binary classification. Among the various optimization techniques of the sigmoid function, Adam optimization technique iteratively updates network weights based on training data. Traditional Adam optimizer fails to converge model within certain epochs when the initial values for parameters are situated at the gentle region of the error surface. The continuous movement of the convergence curve in the direction of history can overshoot the goal and oscillate back and forth incessantly before converging to the global minima. The traditional Adam optimizer with a higher learning rate collapses after several epochs for the high-dimensional dataset. The proposed Improved Adam (iAdam) technique is a combination of the look-ahead mechanism and adaptive learning rate for each parameter. It improves the momentum of traditional Adam by evaluating the gradient after applying the current velocity. iAdam also acts as the correction factor to the momentum of Adam. Further, it works efficiently for the high-dimensional dataset and converges considerably to the smallest error within the specified epochs even at higher learning rates. The proposed technique is compared with several traditional methods which demonstrates that iAdam is suitable for the classification of high-dimensional data and it also prevents the model from overfitting by effectively handling bias-variance trade-offs.

Toward high degree flexible routing in collision-free FMSs through automated guided vehicles’ dynamic strategy: A simulation metamodel

The present study proposes a simulation-based model of high degree flexible routing to prevent collisions and deadlocks in flexible manufacturing systems (FMSs). High degree flexible routing concurrently involves classical routing strategies (specifies the destination and static/dynamic routing between origin and destination) with routing flexibility (considers alternatives to select as destination). A new approach named Tandem-Queue-Link with a look-ahead (TQL-A) is proposed in which the look-ahead mechanism supports deadlock prevention. The TQL-A performs a collision-free status for automated guided vehicles (AGVs) routing with flexible dynamic strategy. The main contribution of the present study is to implement the routing flexible levels (RFLs) by dynamic flexible routing (DFR) as a high degree flexible routing strategy. The most efficient parameters of this highly complex FMS are designed using a multi-objective nonlinear programming model extracted from simulation metamodels based on different hypotheses. Results support some of the considered hypotheses and show the complicated relationship between different values of FMS design parameters and different levels of routing flexibility. Metamodel validity through a typical case study defines that the higher values of velocity and acceleration/deceleration can compensate the level of routing flexibility but other parameters that are influenced by the active parameter (i.e. those related to loading/unloading time) have not any sensible influence on the output.

A Case for Malleable Thread-Level Linear Algebra Libraries: The LU Factorization With Partial Pivoting

We propose two novel techniques for overcoming load-imbalance encountered when implementing so-called look-ahead mechanisms in relevant dense matrix factorizations for the solution of linear systems. Both techniques target the scenario where two thread teams are created/activated during the factorization, with each team in charge of performing an independent task/branch of execution. The first technique promotes worker sharing (WS) between the two tasks, allowing the threads of the task that completes first to be reallocated for use by the costlier task. The second technique allows a fast task to alert the slower task of completion, enforcing the early termination (ET) of the second task, and a smooth transition of the factorization procedure into the next iteration. The two mechanisms are instantiated via a new malleable thread-level implementation of the basic linear algebra subprograms , and their benefits are illustrated via an implementation of the LU factorization with partial pivoting enhanced with look-ahead. Concretely, our experimental results on an Intel-Xeon system with 12 cores show the benefits of combining WS+ET, reporting competitive performance in comparison with a task-parallel runtime-based solution.

Order scheduling with tardiness objective: Improved approximate solutions

The problem addressed in this paper belongs to the topic of order scheduling, in which customer orders – composed of different individual jobs – are scheduled so the objective sought refers to the completion times of the complete orders. Despite the practical and theoretical relevance of this problem, the literature on order scheduling is not very abundant as compared to job scheduling. However, there are several contributions with the objectives of minimising the weighted sum of completion times of the orders, the number of late orders, or the total tardiness of the orders. In this paper, we focus in the last objective, which is known to be NP-hard and for which some constructive heuristics have been proposed. We intend to improve this state-of-the-art regarding approximate solutions by proposing two different methods: Whenever extremely fast (negligible time) solutions are required, we propose a new constructive heuristic that incorporates a look-ahead mechanism to estimate the objective function at the time that the solution is being built. For the scenarios where longer decision intervals are allowed, we propose a novel matheuristic strategy to provide extremely good solutions. The extensive computational experience carried out shows that the two proposals are the most efficient for the indicated scenarios.

Gradient-based pre-processing for intra prediction in High Efficiency Video Coding

In order to reach higher coding efficiency compared to its predecessor, a state-of-the-art video compression standard, the High Efficiency Video Coding (HEVC), has been designed to rely on many improved coding tools and sophisticated techniques. The new features are achieving significant coding efficiency but at the cost of huge implementation complexity. This complexity has increased the HEVC encoders’ need for fast algorithms and hardware friendly implementations. In fact, encoders have to perform the different encoding decisions, overcoming the real-time encoding constraint while taking care of coding efficiency. In this sense, in order to reduce the encoding complexity, HEVC encoders rely on look-ahead mechanisms and pre-processing solutions. In this context, we propose a gradient-based pre-processing stage. We investigate particularly the Prewitt operator used to generate the gradient and we propose necessary approaches that enhance the gradient performance of detecting the HEVC intra modes. We also set different probability scenarios, based on the gradient information, in order to speed up the mode search process. Moreover, we propose a gradient-based estimation of the texture complexity that we use for coding unit decision. Results show that the proposed algorithm achieves a reduction of 42.8% in encoding time with an increase in BD rate of only 1.1%.

A particle swarm optimization and min–max-based workflow scheduling algorithm with QoS satisfaction for service-oriented grids

In service-orientated grids (SOG) environments, grid workflow schedulers play a critical role in providing quality-of-service (QoS) satisfaction for various end users (EUs) with diverse QoS objectives and optimization requirements. The EU requirements are not only many and conflicting, but also involve constraints of various degrees--loose, moderate or tight. However, most of the existing scheduling approaches violate EU constraints in tight situations and suffer inferior QoS optimization results. In this paper, a constraints-aware multi-QoS workflow scheduling strategy is proposed based on particle swarm optimization (PSO) and a proposed look-ahead heuristic (LAPSO) to improve performance in such situations. The algorithm selects the best scheduling solutions based on the proposed constraint-handling strategy. It hybridises PSO with a novel look-ahead mechanism based on a min---max heuristic, which deterministically improves the quality of the best solutions. Extensive simulation experiments have been carried out to evaluate the performance of the proposed approach. The simulation results show that the LAPSO algorithm guarantees satisfaction (0% violation) of the EU constraints even in tight situations. It also outperforms the comparison algorithm, with about 30% increase, in terms of cumulative QoS satisfaction of optimization requirements. In addition, the new scheme significantly reduces the CPU time by about 75% compared to the benchmark algorithm.

New approximate algorithms for the customer order scheduling problem with total completion time objective

In this paper, we study a customer order scheduling problem where a number of orders, composed of several product types, have to be scheduled on a set of parallel machines, each one capable to process a single product type. The objective is to minimise the sum of the completion times of the orders, which is related to the lead time perceived by the customer, and also to the minimisation of the work-in-process. This problem has been previously studied in the literature, and it is known to be NP-hard even for two product types. As a consequence, the interest lies on devising approximate procedures to obtain fast, good performing schedules. Among the different heuristics proposed for the problem, the ECT (Earliest Completion Time) heuristic by Leung et al. [6] has turned to be the most efficient constructive heuristic, yielding excellent results in a wide variety of settings. These authors also propose a tabu search procedure that constitutes the state-of-the-art metaheuristic for the problem. We propose a new constructive heuristic based on a look-ahead mechanism. The computational experience conducted shows that it clearly outperforms ECT, while having both heuristics the same computational complexity. Furthermore, we propose a greedy search algorithm using a specific neighbourhood that outperforms the existing tabu search procedure for different stopping criteria, both in terms of quality of solutions and of required CPU effort.

High throughput graphics processing unit based Fano decoder

The next-generation high speed wireless technologies, such as WirelessHD, bring the concept of several Gigabits per second data communication. However, forward error correction that takes place at the receiver side has become a real computational challenge. So far, only hardware solutions have offered such high-speed convolutional decoding, which could not be achieved by software solutions. Our paper aims to fill this gap and proposes a software level solution offering such multi-Gbps convolutional decoding. For this intend, we use the massively parallel computing power of NVIDIA GPGPUs and CUDA programming environment to implement a solution. As a decoding method, the Fano algorithm is selected for its relatively low memory requirements and computational complexity. A look-ahead mechanism and compact history in the form of circular queue is presented to support such high throughput. Conducted tests showed that our GPU based algorithm can decode up to 9.25Gbps.

A Complexity Based Look-ahead Mechanism for Shop Floor Decision Making

Manufacturing Complexity has been extensively reported in the literature in the ways it can be classified, modelled, assessed and measured. However, the use of complexity measures for shop floor decision making has not been studied exhaustively. The paper demonstrates the use of a complexity measure for shop floor decision making. Knowledge of the transition of shop floor complexity into undesirability can give a correct indicative for any decision maker to intervene and take the necessary corrective actions to bring the system back to its desirable state. The current paper presents an information-theoretic approach based model which considers production scheduling, maintenance and quality functions simultaneously to come up with a complexity quantifier for a shop floor. At any point in time, various environmental factors like machine failures, corrective maintenance time variability, due date alteration, process shifts etc. are simulated into the future and their effect on the job states both in terms of schedule and quality are observed. The probability of these state occurrences is then quantified as complexity of situation which is subsequently scaled to a penalty based “undesirability index”. The complexity is re-evaluated every time a fresh event occurs, thus enabling the decision maker to be more predictive about the future state of the system and in turn, the performance of each of the manufacturing functions. Thus, the model so developed, deploys a complexity based look-ahead mechanism for identifying undesirable state transitions. The novelty of the work lies in the way it has been developed keeping the perspective of a shop floor manager in view and also in the way it can be readily used for decision making.

Cost-Minimizing Dynamic Migration of Content Distribution Services into Hybrid Clouds

With the recent advent of cloud computing technologies, a growing number of content distribution applications are contemplating a switch to cloud-based services, for better scalability and lower cost. Two key tasks are involved for such a move: to migrate the contents to cloud storage, and to distribute the web service load to cloud-based web services. The main issue is to best utilize the cloud as well as the application provider’s existing private cloud, to serve volatile requests with service response time guarantee at all times, while incurring the minimum operational cost. While it may not be too difficult to design a simple heuristic, proposing one with guaranteed cost optimality over a long run of the system constitutes an intimidating challenge. Employing Lyapunov optimization techniques, we design a dynamic control algorithm to optimally place contents and dispatch requests in a hybrid cloud infrastructure spanning geo-distributed data centers, which minimizes overall operational cost over time, subject to service response time constraints. Rigorous analysis shows that the algorithm nicely bounds the response times within the preset QoS target, and guarantees that the overall cost is within a small constant gap from the optimum achieved by a T-slot lookahead mechanism with known future information. We verify the performance of our dynamic algorithm with prototype-based evaluation.

Scaling Social Media Applications Into Geo-Distributed Clouds

Federation of geo-distributed cloud services is a trend in cloud computing that, by spanning multiple data centers at different geographical locations, can provide a cloud platform with much larger capacities. Such a geo-distributed cloud is ideal for supporting large-scale social media applications with dynamic contents and demands. Although promising, its realization presents challenges on how to efficiently store and migrate contents among different cloud sites and how to distribute user requests to the appropriate sites for timely responses at modest costs. These challenges escalate when we consider the persistently increasing contents and volatile user behaviors in a social media application. By exploiting social influences among users, this paper proposes efficient proactive algorithms for dynamic, optimal scaling of a social media application in a geo-distributed cloud. Our key contribution is an online content migration and request distribution algorithm with the following features: 1) future demand prediction by novelly characterizing social influences among the users in a simple but effective epidemic model; 2) one-shot optimal content migration and request distribution based on efficient optimization algorithms to address the predicted demand; and 3) a Δ(t)-step look-ahead mechanism to adjust the one-shot optimization results toward the offline optimum. We verify the effectiveness of our online algorithm by solid theoretical analysis, as well as thorough comparisons to ready algorithms including the ideal offline optimum, using large-scale experiments with dynamic realistic settings on Amazon Elastic Compute Cloud (EC2).

Improving local-search metaheuristics through look-ahead policies

As a basic principle, look-ahead approaches investigate the outcomes of potential future steps to evaluate the quality of alternative search directions. Different policies exist to set up look-ahead methods differing in the object of inspection and in the extensiveness of the search. In this work, two original look-ahead strategies are developed and tested through numerical experiments. The first method introduces a look-ahead mechanism that acts as a hyper-heuristic for comparing and selecting local-search operators. The second method uses a look-ahead strategy on a lower level in order to guide a local-search metaheuristic. The proposed approaches are implemented using a hyper-heuristic framework. They are tested against alternative methods using two different competition benchmarks, including a comparison with results given in literature. Furthermore, in a second set of experiments, a detailed investigation regarding the influence of particular parameter values is executed for one method. The experiments reveal that the inclusion of a simple look-ahead principle into an iterated local-search procedure significantly improves the outcome regarding the considered benchmarks.