On detecting cherry-picked generalizations

Generalizing from detailed data to statements in a broader context is often critical for users to make sense of large data sets. Correspondingly, poorly constructed generalizations might convey misleading information even if the statements are technically supported by the data. For example, a cherry-picked level of aggregation could obscure substantial sub-groups that oppose the generalization. We present a framework for detecting and explaining cherry-picked generalizations by refining aggregate queries. We present a scoring method to indicate the appropriateness of the generalizations. We design efficient algorithms for score computation. For providing a better understanding of the resulting score, we also formulate practical explanation tasks to disclose significant counterexamples and provide better alternatives to the statement. We conduct experiments using real-world data sets and examples to show the effectiveness of our proposed evaluation metric and the efficiency of our algorithmic framework.

Download Full-text

Data segmentation based on the local intrinsic dimension

Scientific Reports ◽

10.1038/s41598-020-72222-0 ◽

2020 ◽

Vol 10 (1) ◽

Author(s):

Michele Allegra ◽

Elena Facco ◽

Francesco Denti ◽

Alessandro Laio ◽

Antonietta Mira

Keyword(s):

High Dimensional Data ◽

Large Data ◽

Large Data Sets ◽

High Dimensional ◽

Data Sets ◽

Imaging Data ◽

Unsupervised Segmentation ◽

Real World Data ◽

Data Set ◽

Intrinsic Dimension

Abstract One of the founding paradigms of machine learning is that a small number of variables is often sufficient to describe high-dimensional data. The minimum number of variables required is called the intrinsic dimension (ID) of the data. Contrary to common intuition, there are cases where the ID varies within the same data set. This fact has been highlighted in technical discussions, but seldom exploited to analyze large data sets and obtain insight into their structure. Here we develop a robust approach to discriminate regions with different local IDs and segment the points accordingly. Our approach is computationally efficient and can be proficiently used even on large data sets. We find that many real-world data sets contain regions with widely heterogeneous dimensions. These regions host points differing in core properties: folded versus unfolded configurations in a protein molecular dynamics trajectory, active versus non-active regions in brain imaging data, and firms with different financial risk in company balance sheets. A simple topological feature, the local ID, is thus sufficient to achieve an unsupervised segmentation of high-dimensional data, complementary to the one given by clustering algorithms.

Download Full-text

An example of spectrum imaging used for comparison of EELS quantitative analysis techniques on Al-Li

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s042482010008794x ◽

1991 ◽

Vol 49 ◽

pp. 726-727

Author(s):

John A. Hunt

Keyword(s):

Quantitative Analysis ◽

Large Data ◽

Difference Spectrum ◽

Large Data Sets ◽

Foil Thickness ◽

Data Sets ◽

Analysis Techniques ◽

Spectrum Imaging ◽

Normal Spectrum ◽

Electron Energy Loss

Spectrum-imaging is a useful technique for comparing different processing methods on very large data sets which are identical for each method. This paper is concerned with comparing methods of electron energy-loss spectroscopy (EELS) quantitative analysis on the Al-Li system. The spectrum-image analyzed here was obtained from an Al-10at%Li foil aged to produce δ' precipitates that can span the foil thickness. Two 1024 channel EELS spectra offset in energy by 1 eV were recorded and stored at each pixel in the 80x80 spectrum-image (25 Mbytes). An energy range of 39-89eV (20 channels/eV) are represented. During processing the spectra are either subtracted to create an artifact corrected difference spectrum, or the energy offset is numerically removed and the spectra are added to create a normal spectrum. The spectrum-images are processed into 2D floating-point images using methods and software described in [1].

Download Full-text

Cluster analysis for large data sets: applications to individual aerosol particles from the mid-pacific

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s0424820100132078 ◽

1992 ◽

Vol 50 (2) ◽

pp. 1488-1489

Author(s):

Thomas W. Shattuck ◽

James R. Anderson ◽

Neil W. Tindale ◽

Peter R. Buseck

Keyword(s):

Cluster Analysis ◽

Chemical Reactivity ◽

Large Data ◽

Large Data Sets ◽

Particle Analysis ◽

Data Sets ◽

Halogen Chemistry ◽

Complete Study ◽

Components Analysis ◽

Automated Scanning

Individual particle analysis involves the study of tens of thousands of particles using automated scanning electron microscopy and elemental analysis by energy-dispersive, x-ray emission spectroscopy (EDS). EDS produces large data sets that must be analyzed using multi-variate statistical techniques. A complete study uses cluster analysis, discriminant analysis, and factor or principal components analysis (PCA). The three techniques are used in the study of particles sampled during the FeLine cruise to the mid-Pacific ocean in the summer of 1990. The mid-Pacific aerosol provides information on long range particle transport, iron deposition, sea salt ageing, and halogen chemistry.Aerosol particle data sets suffer from a number of difficulties for pattern recognition using cluster analysis. There is a great disparity in the number of observations per cluster and the range of the variables in each cluster. The variables are not normally distributed, they are subject to considerable experimental error, and many values are zero, because of finite detection limits. Many of the clusters show considerable overlap, because of natural variability, agglomeration, and chemical reactivity.

Download Full-text

Faculty Opinions recommendation of Detecting novel associations in large data sets.

Faculty Opinions – Post-Publication Peer Review of the Biomedical Literature ◽

10.3410/f.13805958.793484294 ◽

2014 ◽

Author(s):

Daniel Lee

Keyword(s):

Large Data ◽

Large Data Sets ◽

Data Sets ◽

Novel Associations

Download Full-text

NVESTIGATION OF THE EFFICIENCY OF DISTRIBUTED INFORMATION SYSTEMS BASED ON THE PROCESSING OF LARGE AMOUNTS OF DATA

Visnyk Universytetu “Ukraina” ◽

10.36994/2707-4110-2019-2-23-03 ◽

2019 ◽

Author(s):

Mykhajlo Klymash ◽

Olena Hordiichuk — Bublivska ◽

Ihor Tchaikovskyi ◽

Oksana Urikova

Keyword(s):

Distributed Systems ◽

Large Data ◽

Large Data Sets ◽

Data Sets ◽

Data Decomposition ◽

Distributed Information ◽

Software Model ◽

Computing Performance ◽

Mapreduce Model ◽

Singular Data

In this article investigated the features of processing large arrays of information for distributed systems. A method of singular data decomposition is used to reduce the amount of data processed, eliminating redundancy. Dependencies of computational efficiency on distributed systems were obtained using the MPI messaging protocol and MapReduce node interaction software model. Were analyzed the efficiency of the application of each technology for the processing of different sizes of data: Non — distributed systems are inefficient for large volumes of information due to low computing performance. It is proposed to use distributed systems that use the method of singular data decomposition, which will reduce the amount of information processed. The study of systems using the MPI protocol and MapReduce model obtained the dependence of the duration calculations time on the number of processes, which testify to the expediency of using distributed computing when processing large data sets. It is also found that distributed systems using MapReduce model work much more efficiently than MPI, especially with large amounts of data. MPI makes it possible to perform calculations more efficiently for small amounts of information. When increased the data sets, advisable to use the Map Reduce model.

Download Full-text

Using Large Data Sets and Statistical Tools to Define “Failure” and Demonstrate the Safety of DPR

Proceedings of the Water Environment Federation ◽

10.2175/193864718824828128 ◽

2018 ◽

Vol 2018 (6) ◽

pp. 38-39

Author(s):

Austa Parker ◽

Yan Qu ◽

David Hokanson ◽

Jeff Soller ◽

Eric Dickenson ◽

...

Keyword(s):

Large Data ◽

Large Data Sets ◽

Data Sets ◽

Statistical Tools

Download Full-text

Online Judging Platform Utilizing Dynamic Plagiarism Detection Facilities

Computers ◽

10.3390/computers10040047 ◽

2021 ◽

Vol 10 (4) ◽

pp. 47

Author(s):

Fariha Iffath ◽

A. S. M. Kayes ◽

Md. Tahsin Rahman ◽

Jannatul Ferdows ◽

Mohammad Shamsul Arefin ◽

...

Keyword(s):

Source Code ◽

Large Data ◽

Large Data Sets ◽

Detection Technique ◽

Data Sets ◽

Plagiarism Detection ◽

Source Codes ◽

Efficient Detection ◽

Mathematical Problems ◽

Automatic Scoring

A programming contest generally involves the host presenting a set of logical and mathematical problems to the contestants. The contestants are required to write computer programs that are capable of solving these problems. An online judge system is used to automate the judging procedure of the programs that are submitted by the users. Online judges are systems designed for the reliable evaluation of the source codes submitted by the users. Traditional online judging platforms are not ideally suitable for programming labs, as they do not support partial scoring and efficient detection of plagiarized codes. When considering this fact, in this paper, we present an online judging framework that is capable of automatic scoring of codes by detecting plagiarized contents and the level of accuracy of codes efficiently. Our system performs the detection of plagiarism by detecting fingerprints of programs and using the fingerprints to compare them instead of using the whole file. We used winnowing to select fingerprints among k-gram hash values of a source code, which was generated by the Rabin–Karp Algorithm. The proposed system is compared with the existing online judging platforms to show the superiority in terms of time efficiency, correctness, and feature availability. In addition, we evaluated our system by using large data sets and comparing the run time with MOSS, which is the widely used plagiarism detection technique.

Download Full-text