Visual object detection on embedded systems involves a multi-objective optimization problem in the presence of trade-offs between power consumption, processing performance, and detection accuracy. For a new Pareto solution with high processing performance and low power consumption, this paper proposes a hardware architecture for decision tree ensemble using multiple channels of features. For efficient detection, the proposed architecture utilizes the dimensionality of feature channels in addition to parallelism in image space and adopts task scheduling to attain random memory access without conflict. Evaluation results show that an FPGA implementation of the proposed architecture with an aggregated channel features pedestrian detector can process 229 million samples per second at 100MHz operation frequency while it requires a relatively small amount of resources. Consequently, the proposed architecture achieves 350fps processing performance for 1080P Full HD images and outperforms conventional object detection hardware architectures developed for embedded systems.

Multi-task joint sparse representation (MTJSR) is one kind of efficient multi-task learning (MTL) method for solving different problems together using a shared sparse representation. Based on the learning mechanism in human, which is a self-paced learning by gradually training the tasks from easy to difficult, I apply this mechanism into MTJSR, and propose a multi-task joint sparse representation with self-paced learning (MTJSR-SP) algorithm. In MTJSR-SP, the self-paced learning mechanism is considered as a regularizer of optimization function, and an iterative optimization is applied to solve it. Comparing with the traditional MTL methods, MTJSR-SP has more robustness to the noise and outliers. The experimental results on some datasets, i.e. two synthesized datasets, four datasets from UCI machine learning repository, an oxford flower dataset and a Caltech-256 image categorization dataset, are used to validate the efficiency of MTJSR-SP.

We propose a weighted subtask controller and sufficient conditions for boundedness of the controller both velocity and acceleration domain. Prior to designing the subtask controller, a task controller is designed for global asymptotic stability of task space error and subtask error. Although the subtask error converges to zero by the task controller, the boundedness of the subtask controller is also important, therefore its boundedness conditions are presented. The weighted pseudo inverse is introduced to relax the constraints of the null-space of Jacobian. Using the pseudo inverse, we design subtask controller and propose sufficient conditions for boundedness of the auxiliary signal to show the existence of the inverse kinematic solution. The results of experiments using 7-DOF WAM show the effectiveness of the proposed controller.

Reinforcement learning has been used to adaptive service composition. However, traditional algorithms are not suitable for large-scale service composition. Based on Q-Learning algorithm, a multi-task oriented algorithm named multi-Q learning is proposed to realize subtask-assistance strategy for large-scale and adaptive service composition. Differ from previous studies that focus on one task, we take the relationship between multiple service composition tasks into account. We decompose complex service composition task into multiple subtasks according to the graph theory. Different tasks with the same subtasks can assist each other to improve their learning speed. The results of experiments show that our algorithm could obtain faster learning speed obviously than traditional Q-learning algorithm. Compared with multi-agent Q-learning, our algorithm also has faster convergence speed. Moreover, for all involved service composition tasks that have the same subtasks between each other, our algorithm can improve their speed of learning optimal policy simultaneously in real-time.

Recently, mobile applications for recording everyday meals draw much attention for self dietary. However, most of the applications return food calorie values simply associated with the estimated food categories, or need for users to indicate the rough amount of foods manually. In fact, it has not been achieved to estimate food calorie from a food photo with practical accuracy, and it remains an unsolved problem. Then, in this paper, we propose estimating food calorie from a food photo by simultaneous learning of food calories, categories, ingredients and cooking directions using deep learning. Since there exists a strong correlation between food calories and food categories, ingredients and cooking directions information in general, we expect that simultaneous training of them brings performance boosting compared to independent single training. To this end, we use a multi-task CNN. In addition, in this research, we construct two kinds of datasets that is a dataset of calorie-annotated recipe collected from Japanese recipe sites on the Web and a dataset collected from an American recipe site. In the experiments, we trained both multi-task and single-task CNNs, and compared them. As a result, a multi-task CNN achieved the better performance on both food category estimation and food calorie estimation than single-task CNNs. For the Japanese recipe dataset, by introducing a multi-task CNN, 0.039 were improved on the correlation coefficient, while for the American recipe dataset, 0.090 were raised compared to the result by the single-task CNN. In addition, we showed that the proposed multi-task CNN based method outperformed search-based methods proposed before.

This paper addresses the task of assigning labels of fine-grained named entity (NE) types to Wikipedia articles. Information of NE types are useful when extracting knowledge of NEs from natural language text. It is common to apply an approach based on supervised machine learning to named entity classification. However, in a setting of classifying into fine-grained types, one big challenge is how to alleviate the data sparseness problem since one may obtain far fewer instances for each fine-grained types. To address this problem, we propose two methods. First, we introduce a multi-task learning framework, in which NE type classifiers are all jointly trained with a neural network. The neural network has a hidden layer, where we expect that effective combinations of input features are learned across different NE types. Second, we propose to extend the input feature set by exploiting the hyperlink structure of Wikipedia. While most of previous studies are focusing on engineering features from the articles' contents, we observe that the information of the contexts the article is mentioned can also be a useful clue for NE type classification. Concretely, we propose to learn article vectors (i.e. entity embeddings) from Wikipedia's hyperlink structure using a Skip-gram model. Then we incorporate the learned article vectors into the input feature set for NE type classification. To conduct large-scale practical experiments, we created a new dataset containing over 22,000 manually labeled articles. With the dataset, we empirically show that both of our ideas gained their own statistically significant improvement separately in classification accuracy. Moreover, we show that our proposed methods are particularly effective in labeling infrequent NE types. We've made the learned article vectors publicly available. The labeled dataset is available if one contacts the authors.

Hotspots occur frequently in 3D multi-core processors (3D-MCPs), and they may adversely impact both the reliability and lifetime of a system. We present a new thermally constrained task scheduler based on a thermal-pattern-aware voltage assignment (TPAVA) to reduce hotspots in and optimize the performance of 3D-MCPs. By analyzing temperature profiles of different voltage assignments, TPAVA pre-emptively assigns different initial operating-voltage levels to cores for reducing temperature increase in 3D-MCPs. The proposed task scheduler consists of an on-line allocation strategy and a new voltage-scaling strategy. In particular, the proposed on-line allocation strategy uses the temperature-variation rates of the cores and takes into two important thermal behaviors of 3D-MCPs that can effectively minimize occurrences of hotspots in both thermally homogeneous and heterogeneous 3D-MCPs. Furthermore, a new vertical-grouping voltage scaling (VGVS) strategy that considers thermal correlation in 3D-MCPs is used to handle thermal emergencies. Experimental results indicate that, when compared to a previous online thermally constrained task scheduler, the proposed task scheduler can reduce hotspot occurrences by approximately 66% (71%) and improve throughput by approximately 8% (2%) in thermally homogeneous (heterogeneous) 3D-MCPs. These results indicate that the proposed task scheduler is an effective technique for suppressing hotspot occurrences and optimizing throughput for 3D-MCPs subject to thermal constraints.

Improving energy efficiency is critical for embedded systems in our rapidly evolving information society. Near real-time data processing tasks, such as multimedia streaming applications, exhibit a common fact that their deadline periods are longer than their input intervals due to buffering. In general, executing tasks at lower performance is more energy efficient. On the other hand, higher performance is necessary for huge tasks to meet their deadlines. To minimize the energy consumption while meeting deadlines strictly, adaptive task scheduling including dynamic performance mode selection is very important. In this work, we propose an energy efficient slack-based task scheduling algorithm for such tasks by adapting to task size variations and applying DVFS with the help of statistical analysis. We confirmed that our proposal can further reduce the energy consumption when compared to oracle frame-based scheduling.

In this paper, we propose a redundant task allocation method for multi-core systems based on the Duplication with Temporary Triple-Modular Redundancy and Reconfiguration (DTTR) scheme. The proposed method determines task allocation of a given task graph to a given multi-core system model from task scheduling in given fault patterns. Fault patterns defined in this paper consist of a set of faulty cores and a set of surviving cores. To optimize the average failure rate of the system, task scheduling minimizes the execution time of the task graph preserving the property of the DTTR scheme. In addition, we propose a selection method of fault patterns to be scheduled to reduce the task allocation time. In the experiments, at first, we evaluate the proposed selection method of fault patterns in terms of the task allocation time. Then, we compare the average failure rate among the proposed method, a task allocation method which packs tasks into particular cores as much as possible, a task allocation method based on Simulated Annealing (SA), a task allocation method based on Integer Linear Programming (ILP), and a task allocation method based on task scheduling without considering the property of the DTTR scheme. The experimental results show that task allocation by the proposed method results in nearly the same average failure rate by the SA based method with shorter task allocation time.

Task mapping is one of the most important design processes in embedded manycore systems. This paper proposes a static task mapping technique for manycore real-time systems. The technique minimizes the number of cores while satisfying deadline constraints of individual tasks.

Task scheduling is one of the most important processes in the design of multicore computing systems. This paper presents a technique for scheduling of malleable tasks. Our scheduling technique decides not only the execution order of the tasks but also the number of cores assigned to the individual tasks, simultaneously. We formulate the scheduling problem as an integer linear programming (ILP) problem, and the optimal schedule can be obtained by solving the ILP problem. Experiments using a standard task-set suite clarify the strength of this work.

The purpose of this study was to identify the key variables that determine the quality of the auditory environment, for the purposes of workplace auditory design and assessment. To this end, we characterized changes in oscillatory neural activity in electroencephalographic (EEG) data recorded from subjects who performed an intellectual activity while exposed to fluctuating ambient noise. Seven healthy men participated in the study. Subjects performed a verbal and spatial task that used the 3-back task paradigm to study working memory. During the task, subjects were presented with auditory stimuli grouped by increasing high-frequency content: (1) a sound with frequencies similar to Brownian noise and no modulation; (2) an amplitude-modulated sound with frequencies similar to white noise; (3) amplitude-modulated pink noise; and (4) amplitude-modulated Brownian noise. Upon presentation, we observed a characteristic change in three EEG bands: theta (4-8Hz), alpha (8-13Hz), and beta (13-30Hz). In particular, a frequency-dependent enhancement and reduction of power was observed in the theta and beta bands, respectively.

In this paper, a discrete particle swarm optimization method is proposed to solve the multi-objective task assignment problem in distributed environment. The objectives of optimization include the makespan for task execution and the budget caused by resource occupation. A two-stage approach is designed as follows. In the first stage, several artificial particles are added into the initialized swarm to guide the search direction. In the second stage, we redefine the operators of the discrete PSO to implement addition, subtraction and multiplication. Besides, a fuzzy-cost-based elite selection is used to improve the computational efficiency. Evaluation shows that the proposed algorithm achieves Pareto improvement in comparison to the state-of-the-art algorithms.

Multi-task learning in deep neural networks has been proven to be effective for acoustic modeling in speech recognition. In the paper, this technique is applied to Mandarin-English code-mixing recognition. For the primary task of the senone classification, three schemes of the auxiliary tasks are proposed to introduce the language information to networks and improve the prediction of language switching. On the real-world Mandarin-English test corpus in mobile voice search, the proposed schemes enhanced the recognition on both languages and reduced the relative overall error rates by 3.5%, 3.8% and 5.8% respectively.

For an efficient processing of large data in a distributed system, Hadoop MapReduce performs task scheduling such that tasks are distributed with consideration of the data locality. The data locality, however, is limitedly exploited, since it is pursued one node at a time basis without considering the global optimality. In this paper, we propose a novel task scheduling algorithm that globally considers the data locality. Through experiments, we show our algorithm improves the performance of MapReduce in various situations.

Dynamic Voltage/Frequency Scaling (DVFS) allows designers to improve energy efficiency through adjusting supply voltage at runtime in order to meet the workload demand. Previous works solving real-time DVFS problems often refer to the canonical schedules with the exponential length. Other solutions for online scheduling depend on empirical or stochastic heuristics, which potentially result in frequent fluctuations of voltage/speed scaling. This paper aims at increasing the schedule predictability using period transformation in the pinwheel task model and improves the control on power-awareness by decreasing the speeds of as many tasks as possible to the same level. Experimental results show the maximum energy savings of 6% over the recent Dynamic Power Management (DPM) method and 12% over other slack reclamation algorithms.

Recently, due to the development of design and manufacturing technologies for VLSI systems, an embedded system becomes more and more complex. Consequently, not only the performance of chips, but also the flexibility and dynamic adaptation of the implemented systems are required. To achieve these requirements, a partially reconfigurable device is promising. In this paper, we propose an efficient data structure to manage the reconfigurable units. And then, on the assumption that each task utilizes the rectangle shaped resources, a very simple MER enumeration algorithm based on this data structure is proposed. By utilizing the result of MER enumeration, the free space on the reconfigurable device can be used sufficiently. We analyze the complexity of the proposed algorithm and confirm its efficiency by experiments.

Operating swarm robots has the virtues of improved performance, fault tolerance, distributed sensing, and so on. The problem is, high overall system costs are the main barrier in managing a system of foraging swarm robots. Moreover, its control algorithm should be scalable and reliable as the foraging (search) spaces become wider. This paper analyzes a nature-inspired cooperative method to reduce the operating costs of the foraging swarm robots through simulation experiments. The aim of this research is to improve efficiency of mechanisms for reducing the cost by developing a new algorithm for the synergistic cooperation of the group. In this paper, we set the evaluation index of energy efficiency considering that the mission success rate as well as energy saving is important. The value is calculated as the number of successful operations against the total consumption of energy in order to also guarantee optimized for the work processing power than the one simple goal of energy savings. The method employs a behavioral model of a honey bee swarm to improve the energy efficiency in collecting crops or minerals. Experiments demonstrate the effectiveness of the approach. The experiment is set a number of strategies to combine the techniques to the proposed and conventional methods. Considering variables such as the area of search space and the size of a swarm, the efficiency comparison test is performed. As the result, the proposed method showed the enhanced energy efficiency of the average 76.9% as compared to the conventional simple model that means reduction of the recharging cost more than 40%.

Recently, notable improvements in voice activity detection (VAD) problem have been achieved by adopting several machine learning techniques. Among them, the deep neural network (DNN) which learns the mapping between the noisy speech features and the corresponding voice activity status with its deep hidden structure has been one of the most popular techniques. In this letter, we propose a novel approach which enhances the robustness of DNN in mismatched noise conditions with multi-task learning (MTL) framework. In the proposed algorithm, a feature enhancement task for speech features is jointly trained with the conventional VAD task. The experimental results show that the DNN with the proposed framework outperforms the conventional DNN-based VAD algorithm.

MapReduce and its open software implementation Hadoop are now widely deployed for big data analysis. As MapReduce runs over a cluster of massive machines, data transfer often becomes a bottleneck in job processing. In this paper, we explore the influence of data transfer to job processing performance and analyze the mechanism of job performance deterioration caused by data transfer oriented congestion at disk I/O and/or network I/O. Based on this analysis, we update Hadoop's Heartbeat messages to contain the real time system status for each machine, like disk I/O and link usage rate. This enhancement makes Hadoop's scheduler be aware of each machine's workload and make more accurate decision of scheduling. The experiment has been done to evaluate the effectiveness of enhanced scheduling methods and discussions are provided to compare the several proposed scheduling policies.