Description
1. Summary
This assignment requires to thoroughly review one high-quality research paper from famous
conferences or journals. You should select one interested paper from the given paper list and
summarize your review into a file.
2. Description
Reading a good research paper can greatly improve your perspective and interests for one
particular research area. Here list several papers that are from high qualified or famous conferences,
and each represents one excellent insight or improvement in a particular real problem.
Please only
select one paper from the below list:
• Taiji: managing global user traffic for large-scale internet services at the edge, SOSP’19.
• EdgeWise: A Better Stream Processing Engine for the Edge, ATC’19.
• Optimizing data-intensive computations in existing libraries with split annotations,
SOSP’19
• StreamBox-TZ: Secure Stream Analytics at the Edge with TrustZone, ATC’19
• Tangram: Bridging Immutable and Mutable Abstractions for Distributed Data Analytics,
ATC’19
• SR3: Customizable Recovery for Stateful Stream Processing Systems, Middleware’ 20
• HiveD: Sharing a GPU Cluster for Deep Learning with Guarantees, OSDI’20
• GrandSLAm: Guaranteeing SLAs for Jobs in Microservices Execution Frameworks,
Eurosys’19
• Efficient, Consistent Distributed Computation with Predictive Treaties, Eurosys’19
• GraphBolt: Dependency-Driven Synchronous Processing of Streaming Graphs, Eurosys’19
• Twine: A Unified Cluster Management System for Shared Infrastructure, OSDI 20
• Ansor: Generating High-Performance Tensor Programs for Deep Learning, OSDI 20
• PipeSwitch: Fast Pipelined Context Switching for Deep Learning Applications, OSDI 20
• Unearthing inter-job dependencies for better cluster scheduling, OSDI 20
• KungFu: Making Training in Distributed Machine Learning Adaptive, OSDI 20
How to Get a Copy of Paper:
Once you’ve found the citation for a paper that is relevant to your advanced science project, the
next step is getting a copy so that you can read it. Some search engines provide links to free
online versions of the paper, if one exists.
There are several ways to find the paper or copies:
• Look for a free online version. Try searching for the full title of the paper in a regular search
engine like Google Scholar, Yahoo, or Microsoft Scholar. The paper may come up multiple
times, and one of those might be a free, downloadable copy. So, if the first link isn’t
downloadable, try another. Tips: For Google Scholar, click the “All # versions”, you may find
some versions can be freely accessed.
• Search directly for the homepage of the first or last author of the paper and see if he or she
has a PDF of the paper on his or her website. If so, you can download it directly from there.
Generally, it is only worth looking up the first author (the one who contributed the most to the
paper) or the last author (usually the professor in whose lab the work was done and who
supervised the science project).
• Go directly to the online homepage of the journal in which the paper was published. Some
scientific journals are “open-source,” meaning that their content is always free online to the
public. Others are free online (often after registering with the website) if the paper
• Utilize the CSULB network. You can utilize the “CSULB Library” (the link is in the above
table) to access the articles with your student account.
Reading Instructions:
Review the paper should concentrate on the motivation and the design methodology of the
proposed work. You should pay attention to the author’s motivation, design novelty, proposed
system/architecture, and comparison with previous work. You may skip the technical details that
you are not familiar with. You should focus on the improvement or novel part of the proposed
work.
2. Write Instructions
You should write the summary via Word or Latex. The submitted file must be a pdf format.
Component 1: Summary of paper
In this component, you should use one or two paragraphs to summarize the paper your read. You
should introduce the problem they provided, the design motivation, design novelty, proposed
architecture/methodology, and evaluated results.
Component 2: Your views or analysis
In this component, you should provide your analysis or perspectives about the paper. You should
comment on their proposed problem, discuss their design, and provide your own views about their
work. For example, you can point out the possible improvement of the proposed solution, or clarify
the drawback or inefficiency of their work from any point. You can also analyze the difference
from previous work, and show your perspective of their novelty, or classify the possible
shortcoming in their design. This component will be the majority part of your report.
Formats requirements:
• We will do similarity check. The accepted similarity must be less than 20%, otherwise it
will be taken as plagiarism.
• The length of your report is around 2 pages. (Except the title and name info, etc.)
• The font should be Times New Roman, Size 11, single line space, 0 bt in before or after
space, 8.5*11’.
• The submission only includes one pdf file, no other files should be submitted.
3. Grades Criteria:
Details Points
Submitted file with correct format. 10 pts
Typo identification. 10 pts
Detailed paper summary. 20 pts
Details of your analysis or discussion. 60 pts
