arXiv subjects

Ehsan Kazemi

Publications and source records attributed to Ehsan Kazemi.

24 records · Page 2Linked to original sources

arXiv2018

Data Summarization at Scale: A Two-Stage Submodular Approach

The sheer scale of modern datasets has resulted in a dire need for summarization techniques that identify representative elements in a dataset. Fortunately, the vast majority of data summarization tasks satisfy an intuitive diminishing returns condition known as submodularity, which allows us to find nearly-optimal solutions in linear time. We focus on a two-stage submodular framework where the goal is to use some given training functions to reduce the ground set so that optimizing new functions (drawn from the same distribution) over the reduced set provides almost as much value as optimizing them over the entire ground set. In this paper, we develop the first streaming and distributed solutions to this problem. In addition to providing strong theoretical guarantees, we demonstrate both the utility and efficiency of our algorithms on real-world tasks including image summarization and ride-share optimization.

cs.LG↗

arXiv2018

Comparison Based Learning from Weak Oracles

There is increasing interest in learning algorithms that involve interaction between human and machine. Comparison-based queries are among the most natural ways to get feedback from humans. A challenge in designing comparison-based interactive learning algorithms is coping with noisy answers. The most common fix is to submit a query several times, but this is not applicable in many situations due to its prohibitive cost and due to the unrealistic assumption of independent noise in different repetitions of the same query. In this paper, we introduce a new weak oracle model, where a non-malicious user responds to a pairwise comparison query only when she is quite sure about the answer. This model is able to mimic the behavior of a human in noise-prone regions. We also consider the application of this weak oracle model to the problem of content search (a variant of the nearest neighbor search problem) through comparisons. More specifically, we aim at devising efficient algorithms to locate a target object in a database equipped with a dissimilarity metric via invocation of the weak comparison oracle. We propose two algorithms termed WORCS-I and WORCS-II (Weak-Oracle Comparison-based Search), which provably locate the target object in a number of comparisons close to the entropy of the target distribution. While WORCS-I provides better theoretical guarantees, WORCS-II is applicable to more technically challenging scenarios where the algorithm has limited access to the ranking dissimilarity between objects. A series of experiments validate the performance of our proposed algorithms.

cs.LG↗

arXiv2018

Do Less, Get More: Streaming Submodular Maximization with Subsampling

In this paper, we develop the first one-pass streaming algorithm for submodular maximization that does not evaluate the entire stream even once. By carefully subsampling each element of data stream, our algorithm enjoys the tightest approximation guarantees in various settings while having the smallest memory footprint and requiring the lowest number of function evaluations. More specifically, for a monotone submodular function and a $p$-matchoid constraint, our randomized algorithm achieves a $4p$ approximation ratio (in expectation) with $O(k)$ memory and $O(km/p)$ queries per element ($k$ is the size of the largest feasible solution and $m$ is the number of matroids used to define the constraint). For the non-monotone case, our approximation ratio increases only slightly to $4p+2-o(1)$. To the best or our knowledge, our algorithm is the first that combines the benefits of streaming and subsampling in a novel way in order to truly scale submodular maximization to massive machine learning problems. To showcase its practicality, we empirically evaluated the performance of our algorithm on a video summarization application and observed that it outperforms the state-of-the-art algorithm by up to fifty fold, while maintaining practically the same utility.

cs.LG↗

arXiv2017

Deletion-Robust Submodular Maximization at Scale

Can we efficiently extract useful information from a large user-generated dataset while protecting the privacy of the users and/or ensuring fairness in representation. We cast this problem as an instance of a deletion-robust submodular maximization where part of the data may be deleted due to privacy concerns or fairness criteria. We propose the first memory-efficient centralized, streaming, and distributed methods with constant-factor approximation guarantees against any number of adversarial deletions. We extensively evaluate the performance of our algorithms against prior state-of-the-art on real-world applications, including (i) Uber-pick up locations with location privacy constraints; (ii) feature selection with fairness constraints for income prediction and crime rate prediction; and (iii) robust to deletion summarization of census data, consisting of 2,458,285 feature vectors.

cs.LG↗

arXiv2016

On the Structure and Efficient Computation of IsoRank Node Similarities

The alignment of protein-protein interaction (PPI) networks has many applications, such as the detection of conserved biological network motifs, the prediction of protein interactions, and the reconstruction of phylogenetic trees [1, 2, 3]. IsoRank is one of the first global network alignment algorithms [4, 5, 6], where the goal is to match all (or most) of the nodes of two PPI networks. The IsoRank algorithm first computes a pairwise node similarity metric, and then generates a matching between the two node sets based on this metric. The metric is a convex combination of a structural similarity score (with weight $ α$) and an extraneous amino-acid sequence similarity score for two proteins (with weight $ 1 - α$). In this short paper, we make two contributions. First, we show that when IsoRank similarity depends only on network structure ($α= 1$), the similarity of two nodes is only a function of their degrees. In other words, IsoRank similarity is invariant to any network rewiring that does not affect the node degrees. This result suggests a reason for the poor performance of IsoRank in structure-only ($ α= 1 $) alignment. Second, using ideas from [7, 8], we develop an approximation algorithm that outperforms IsoRank (including recent versions with better scaling, e.g., [9]) by several orders of magnitude in time and memory complexity, despite only a negligible loss in precision.

q-bio.MN↗

arXiv2013

Mitigating Epidemics through Mobile Micro-measures

Epidemics of infectious diseases are among the largest threats to the quality of life and the economic and social well-being of developing countries. The arsenal of measures against such epidemics is well-established, but costly and insufficient to mitigate their impact. In this paper, we argue that mobile technology adds a powerful weapon to this arsenal, because (a) mobile devices endow us with the unprecedented ability to measure and model the detailed behavioral patterns of the affected population, and (b) they enable the delivery of personalized behavioral recommendations to individuals in real time. We combine these two ideas and propose several strategies to generate such recommendations from mobility patterns. The goal of each strategy is a large reduction in infections, with a small impact on the normal course of daily life. We evaluate these strategies over the Orange D4D dataset and show the benefit of mobile micro-measures, even if only a fraction of the population participates. These preliminary results demonstrate the potential of mobile technology to complement other measures like vaccination and quarantines against disease epidemics.

cs.SI↗