Study proposes a clustering and logistic regression algorithm for PU classification under Non-SCAR.
problem PU classification under Non-SCAR condition when SCAR condition is unsatisfied.
method 2-means clustering followed by logistic regression.
result Efficacy of the proposed algorithm demonstrated on 11 real data sets and a synthetic set.
Generic scarring occurs along stable minimal hypersurfaces in 3-7 dimensional manifolds.
problem Understanding scarring behavior of minimal hypersurfaces along stable ones.
method Analyzing a generic metric on a manifold to show scarring of minimal hypersurfaces.
result Closed, embedded minimal hypersurfaces scarring along stable ones, with diverging area and Morse index.
Minimal hypersurfaces scarring along a fixed one in certain manifolds.
problem Scarring of minimal hypersurfaces in specific manifolds.
method Generic scarring phenomenon for minimal hypersurfaces in thick-at-infinity manifolds with thin foliation.
result Existence of sequences of minimal hypersurfaces scarring along a fixed one, with diverging area and renormalized convergence to the fixed hypersurface.
Test verifies if data meets SCAR assumption for PU learning.
problem Verify if labeling mechanism follows SCAR assumption in PU learning.
method Generate artificial labels, mimic distribution of test statistic.
result Test detects deviations from SCAR and controls type I error.
Study on random surfaces in hyperbolic 3-manifolds, focusing on geometric and topological properties.
problem Distribution of nearly geodesic surfaces in hyperbolic 3-manifolds.
method Invariant measures on the Grassmann bundle G(M) derived from limits of random minimal surfaces.
result Topological limiting measures are totally scarring if M contains a totally geodesic subsurface, while geometrical limiting measures are not.
New model tackles PU data with better accuracy.
problem Addressing positive and unlabeled data challenges.
method Double Exponential Tilting Model (DETM)
result DETM effectively handles selected at random PU data.
Machine learning (ML) training algorithms often possess an inherent self-correcting behavior due to their iterative-convergent nature. Recent systems exploit this property to achieve adaptability and efficiency in unreliable computing environments by relaxing the consistency of execution and allowing calculation errors…
In binary classification, Learning from Positive and Unlabeled data (LePU) is semi-supervised learning but with labeled elements from only one class. Most of the research on LePU relies on some form of independence between the selection process of annotated examples and the features of the annotated class, known as the…
Paper proposes LC-Checkpoint for efficient deep learning model checkpoints.
problem Efficient construction of checkpoints for deep learning models.
method Lossy compression scheme using quantization and priority promotion with Huffman coding.
result LC-Checkpoint achieves up to 28x compression and 5.77x speedup over SCAR.
Improved classifier for PU data using logistic regression.
problem Analysis of Positive Unlabeled data under SCAR assumption.
method Fitting misspecified logistic regression model to PU data.
result The classifier performs on par or better than competitors on real data sets.
Proposes methods to estimate posterior probability and propensity score functions without assuming constant propensity score.
problem Learning from biased positive-unlabeled data.
method Parametric approach to joint estimation of posterior probability and propensity score functions using maximum likelihood and alternating maximization.
result Proposed methods are comparable or better than existing methods based on Expectation-Maximisation scheme.