Provably Efficient Online Nonclairvoyant Adaptive Scheduling

He, Yuxiong; Hsu, Wen-Jing; Leiserson, Charles E.

doi:10.1109/tpds.2008.39

Cited by 28 publications

(4 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We selected sensitivity precision, F‐score, and G‐mean for evaluating the performance. These metrics are widely used in imbalanced learning 78‐82 . All parameters of sensitivity (Equation ) precision (Equation ), sensitivity (Equation ), F‐score (Equation ), and G‐mean (Equation ) are taken from confusion matrix Table 5 and description are given in Table 6.…”

Section: Methodsmentioning

confidence: 99%

“…These metrics are widely used in imbalanced learning. [78][79][80][81][82] All parameters of sensitivity (Equation 31) precision (Equation 30), sensitivity (Equation 31), F-score (Equation 32), and G-mean (Equation 33) are taken from confusion matrix Table 5 and description are given in Table 6. Our main focusing metric is to reduce information loss and excessive elimination with significant improvement in sensitivity and G mean.…”

Section: Evaluation Metricsmentioning

confidence: 99%

See 1 more Smart Citation

Entropy and improved k‐nearest neighbor search based under‐sampling (ENU) method to handle class overlap in imbalanced datasets

Kumar,

Singh,

Yadav

2023

Concurrency and Computation

View full text Add to dashboard Cite

SummaryMany real‐world application datasets such as medical diagnostics, fraud detection, biological classification, risk analysis and so forth are facing class imbalance and overlapping problems. It seriously affects the learning of the classification model on these datasets because minority instances are not visible to the learner in the overlapped region and the performance of learners is biased towards the majority. Undersampling‐based methods are the most commonly used techniques to handle the above‐mentioned problems. The major problem with these methods is excessive elimination and information loss, that is, unable to retain potential informative majority instances. We propose a novel entropy and neighborhood‐based undersampling (ENU) that removed only those majority instances from the overlapped region which are having less informativeness (entropy) score than the threshold entropy. Most of such existing methods improved sensitivity scores significantly but not in many other performance contexts. ENU first computes entropy and threshold score for majority instances and, a local density‐based improved KNN search is used to identify overlapped majority instances. To tackle the problem effectively ENU defined four improved KNN‐based procedures (ENUB, ENUT, ENUC, and ENUR) for effective undersampling. ENU outperformed in sensitivity, G‐mean, and F1‐score average ranking with reduced information loss as compared to the existing state‐of‐the‐art methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Evaluation Metricsmentioning

confidence: 99%

Entropy and improved k‐nearest neighbor search based under‐sampling (ENU) method to handle class overlap in imbalanced datasets

Kumar,

Singh,

Yadav

2023

Concurrency and Computation

View full text Add to dashboard Cite

show abstract

“…Most works related scarcity of samples rely on a base set with adequate samples and pay more attention to few-shot classification rather than learning a general representation under data with long-tail distribution. In addition, some classical strategies [19], [20] against imbalance problem have been made full use in current deep learning frameworks, including resampling [21], [22] and cost-sentitive learning [23], [24]. However, there are still many limitations such as discarding samples for under-sampling or introducing additional noises (e.g., assigning larger penalty on outlier samples).…”

Section: Introductionmentioning

confidence: 99%

Multi-Agent Semi-Siamese Training for Long-tail and Shallow Face Learning

Shi¹,

Zeng²,

Tai³

et al. 2021

Preprint

View full text Add to dashboard Cite

With the recent development of deep convolutional neural networks and large-scale datasets, deep face recognition has made remarkable progress and been widely used in various applications. However, unlike the existing public face datasets, in many real-world scenarios of face recognition, the depth of training dataset is shallow, which means only two face images are available for each ID. With the non-uniform increase of samples, such issue is converted to a more general case, a.k.a long-tail face learning, which suffers from data imbalance and intra-class diversity dearth simultaneously. These adverse conditions damage the training and result in the decline of model performance. Based on the Semi-Siamese Training (SST), we introduce an advanced solution, named Multi-Agent Semi-Siamese Training (MASST), to address these problems. MASST includes a probe network and multiple gallery agents, the former aims to encode the probe features, and the latter constitutes a stack of networks that encode the prototypes (gallery features). For each training iteration, the gallery network, which is sequentially rotated from the stack, and the probe network form a pair of semisiamese networks. We give theoretical and empirical analysis that, given the long-tail (or shallow) data and training loss, MASST smooths the loss landscape and satisfies the Lipschitz continuity with the help of multiple agents and the updating gallery queue. The proposed method is out of extra-dependency, thus can be easily integrated with the existing loss functions and network architectures. It is worth noting that, although multiple gallery agents are employed for training, only the probe network is needed for inference, without increasing the inference cost. Extensive experiments and comparisons demonstrate the advantages of MASST for long-tail and shallow face learning.

show abstract

“…Moreover, unlike the non-clairvoyant algorithms MultiLaps and N-EQUI, both of which require non-uniform speed scaling for an individual job, U-CEQ only requires allocating processors of uniform speed to a job. Thus, in situations where the instantaneous parallelism of a job does not change frequently and can be effectively measured, e.g., by using feedback mechanisms [1,28,45], our IP-clairvoyant algorithm may be easier to implement and more practical.…”

Section: Introductionmentioning

confidence: 99%

Energy-efficient multiprocessor scheduling for flow time and makespan

Sun

Hsu

et al. 2014

Theoretical Computer Science

Self Cite

View full text Add to dashboard Cite

We consider energy-efficient scheduling on multiprocessors, where the speed of each processor can be individually scaled, and a processor consumes power s α when running at speed s, for α > 1. A scheduling algorithm needs to decide at any time both processor allocations and processor speeds for a set of parallel jobs with time-varying parallelism. The objective is to minimize the sum of the total energy consumption and certain performance metric, which in this paper includes total flow time and makespan. For both objectives, we present instantaneous parallelism-clairvoyant (IP-clairvoyant) algorithms that are aware of the instantaneous parallelism of the jobs at any time but not their future characteristics, such as remaining parallelism and work. For total flow time plus energy, we present an O(1)-competitive algorithm, which significantly improves upon the best known non-clairvoyant algorithm and is the first constant competitive result on multiprocessor speed scaling for parallel jobs. In the case of makespan plus energy, which is considered for the first time in the literature, we present an O(ln 1−1/α P )competitive algorithm, where P is the total number of processors. We show that this algorithm is asymptotically optimal by providing a matching lower bound. In addition, we also study non-clairvoyant scheduling for total flow time plus energy, and present an algorithm that achieves O(ln P )-competitive for jobs with arbitrary release time and O(ln 1/α P )-competitive for jobs with identical release time. Finally, we prove an Ω(ln 1/α P ) lower bound on the competitive ratio of any non-clairvoyant algorithm, matching the upper bound of our algorithm for jobs with identical release time.

show abstract

Provably Efficient Online Nonclairvoyant Adaptive Scheduling

Cited by 28 publications

References 43 publications

Entropy and improved k‐nearest neighbor search based under‐sampling (ENU) method to handle class overlap in imbalanced datasets

Entropy and improved k‐nearest neighbor search based under‐sampling (ENU) method to handle class overlap in imbalanced datasets

Multi-Agent Semi-Siamese Training for Long-tail and Shallow Face Learning

Energy-efficient multiprocessor scheduling for flow time and makespan

Contact Info

Product

Resources

About