1,720,973 research outputs found
Going Beyond Counting First Authors in Author Co-citation Analysis
Get PDFThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Evaluating CGAN Frameworks to Generate Performance Data
No full textOne of the most common challenges in implementing machine learning models, where performance is almost entirely dependent on the quantity and quality of the data available, is insufficient data. More so in the field of High-Performance Computing (HPC) where data collection is a long and a nontrivial issue. As a result, data augmentation using generative models [1] has seen a rise in the past few years. It has played a crucial role in improving the performance and accuracy of machine learning models by scaling up the data size without employing exhausting and costly processes.
A lot of the work done within this field is concentrated on image classification and segmentation. With its challenges, data augmentation has also been used in Natural Language Processing [2]. Later works on generative models gave rise data augmentation techniques like Generative Adversarial Networks (GANs) [3], followed by specialized versions like the styleGAN [2], DCGAN [3], etc. for image generation and CGAN (Conditional GAN) [4] and CTGAN [5] for regression. Our work evaluates such generative modeling frameworks in their ability of generating scientific data, specifically performance data. We also evaluate the effectiveness of these methods using open-source Machine Learning data repositories with regression datasets.Computer Scienc
Out-of-Core Graph Coloring Algorithm
No full textOut-of-core algorithms can process data sets that are too large to fit entirely into the computer’s main memory. This thesis develops an out-of-core algorithm for graph coloring. It dynamically partitions the graph into subgraphs, processes them in sequence, and records the color information needed by later subgraphs in a dense format. The algorithm is guaranteed to produce the same coloring as the first-fit in-core algorithm. It employs a new method to compactly record information and automatically resizes the associated data structure to save memory. As there are no pre-existing out-of-core graph coloring codes, the implementation can only be compared to leading in-core graph coloring codes. Based on the geometric mean over 18 graphs from various domains, JP-D1 is 25% faster and uses 13% fewer colors. FirstFit and Boost both use the same number of colors as the presented implementation, but FirstFit is 4 times faster whereas Boost is 6 times slower.Computer Scienc
Variations on the Author
Get PDF“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Appropriate Similarity Measures for Author Cocitation Analysis
Get PDFWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Enabling Transfer Learning Across Heterogeneous Domains Using Input Alignment
No full textAs High Performance Computing (HPC) architectures evolve quickly, gathering data on each architecture to create predictive performance models can be time-consuming. Instead, leveraging existing performance models for predicting the performance of a new application or an existing application on a new architecture is a better use of scientists’ time. However, such knowledge transfer in HPC can be difficult as a result of domain heterogeneity resulting from (1) data distribution shifts due to differences in architectures and (2) incomparable feature space due to differences in the names and numbers of features. Although existing transfer learning techniques can handle the first scenario (distribution shift), the second is unique to HPC and requires attention. We propose a few-shot-learning-based test-time adaptation method for Neural Network (NN)s to address this gap. The proposed approach aligns target features using a fully-connected NN to leverage existing knowledge from a pre-trained source model, and then stacking the results from the source model with a secondary model to capture the uncaptured relationships between the unique target features and labels. Our evaluations with both HPC datasets and Machine Learning (ML) benchmarks demonstrate that this approach can adapt an existing source model into a high-fidelity target model with only a few target samples. Additionally, we propose a novel distance measure to easily quantify the dissimilarity between the source and target datasets, which can help identify and explain the best model to use for knowledge transfer. Our extensive evaluations demonstrate that the proposed few-shot-learning-based test-time adaptation method augmented with input alignment achieves the lowest Mean-Square Error (MSE) when predicting a significantly different target from the source data.Computer Scienc
Enhancing Energy-Efficiency in Cloud Computing and Green Software Design through Accurate Power Measurement for Heterogeneous Hardware
No full textThe growing energy demands of modern computing, from software development to expansive cloud platforms, necessitate accurate power measurement tools to optimize energy usage and address environmental concerns. Although a variety of frameworks and methodologies have been developed to measure power consumption of software, respective inherent limitations of each of these methods make those unsuitable and, in many cases, inapplicable for deployment environments. Technologies relying on methods like Thermal Design Power (TDP) fall short due to their large inaccuracies. Whereas Running Average Power Limit (RAPL) is restricted by its administrative access requirements despite its high precision [1]. It is leaving developers, users, and software-based service providers without practical means to measure or optimize energy usage. To bridge such technology gaps, we leverage ML-based methodologies to propose precise and convenient real-time power consumption estimation technology which is capable of performing irrespective of deployment environments or host specifications. We have taken a novel approach of combining the decision tree algorithm with a multi-variable regression model to adapt to the host environment and capture power trends with precision in the highly heterogeneous computer ecosystems to-date. Our proposed GreenCode-PowerMeter (GCPM) is integrated within the popular platform of VS Code to let its 14 million users estimate power consumption across diverse programming languages and hardware configurations directly within the IDE with a mean absolute percentage error of only 0.075%. This enables developers to adopt energy-efficient coding practices without the complexity of accessing administrative privileges.
Meanwhile, cloud computing's rapid expansion since the end of the past century has raised significant energy consumption and environmental concerns. Accurate power measurement is crucial in addressing such concerns, as precise data helps both cloud providers and consumers to optimize usage by enabling informed resource allocation to meet sustainability goals. Complementing the effectiveness of GreenCode-PowerMeter, we extend the ML-based approach for addressing the challenges of providing an accurate power consumption method in diverse cloud environments where administrative access is unattainable. We propose Cloud Power Meter (CPM) which is trained on 34 cloud instances from providers like AWS, Microsoft Azure, and Google Cloud Platform. CPM achieves a mean absolute percentage error of 0.31% while tested on the dataset that is constructed from the real-world cloud processors’ power measurements based on the CPU benchmarks from SPEC (Standard Performance Evaluation Corporation). The decision tree guarantees performance capacity across heterogeneous cloud environments by locating the ‘CPM Instance’. To demonstrate its efficacy, we perform and present an analysis on 2.7 million Virtual Machines (VMs) traces on Microsoft Azure, identifying a significant 25,000 kWh energy-saving scopes from 17% inefficient allocations in a month.
Both GCPM and CPM outperform the widely deployed TDP-based methods and closely align with the RAPL. Together, GCPM and CPM provide scalable, accurate, and accessible solutions for power measurement in modern computing environments. By addressing inefficiencies and enabling informed decision-making, these tools empower developers and cloud providers to optimize resource allocation, reduce carbon footprints, and contribute to sustainable computing practices.Computer Scienc
Towards Constructing a Tunable Register Allocator
No full textCompilers are the unseen heroes of Computer Science. They apply transformations to source code that improve the performance of applications so that users may use them without fault. The register allocation process is a critical phase of the compiler, assigning program values to fast access storage known as registers. Efficient register allocation can address memory wall challenges, but it is difficult given the task of assigning a finite number of registers to an infinite number of variables in a program, a problem that is NP-Complete. This is further complicated due to varying allocation policies that exist between different architectures. The lack of runtime and system information also makes efficient register allocation challenging. Although work has been done to improve the amount of information available during the register allocation process, it can still be limiting by what information is available in general. Documentation can also be sparse beyond developers choosing to explore a compiler’s source code to understand how it works, let alone the register allocation procedure.
Our research explores ways to create a tunable register allocator. We detail a general overview of the current state of compiler implementations. We explain how one popular compiler toolchain technology, LLVM, performs this, along with a detailed description of its register allocation process. We then investigate ways to expose tunable parameters at the live range level for better control of the register allocation process. Using hardware performance counters, we can provide feedback to the register allocator at the instructions level. We see how learning about certain hardware counters, such as cache miss events, on a variable reference in a baseline allocation can help the compiler develop a better allocation policy. We finally analyze different performance monitoring metrics and how these may be integrated with the register allocator implemented by LLVM.Computer Scienc
Novel Representation Learning Technique Using Graphs For Performance Analytics
No full textMost publicly available datasets are in a tabular format. It is one of the common data formats used for machine learning (ML) applications, especially in HPC, where models solve regression problems, such as predicting the execution time. Existing ML techniques leverage the correlations among features given tabular datasets, disregarding any relationship between the samples. Moreover, the success of the downstream analysis techniques depends on how well information is extracted from the raw features. For high-quality embeddings, existing methods rely on extensive feature engineering and preprocessing steps, which come at a high cost and require a human in the loop. To fill these two gaps, we propose a novel idea of transforming performance data into graphs to leverage the advancement of graph neural network-based (GNN) techniques in capturing complex relationships between features and samples. In contrast to other ML application domains such as social networks, the graph is not given; instead, we need to build it. To address this gap, we propose graph building methods where nodes represent samples, and the edges are automatically inferred iteratively based on the similarity between the features in the samples. We evaluate the effectiveness of the generated embeddings from GNNs based on how well they make even a simple feed-forward neural network perform for regression tasks compared to other state-of-the-art representation learning techniques. Our evaluation demonstrates that even with up to 25% random missing values for each dataset, our method outperforms commonly used graph and deep neural network (DNN)-based approaches and achieves up to 51.77% improvement in MSE loss over the DNN baseline.Computer Scienc
- …
