Search Results for author: Michael Mathioudakis

Found 13 papers, 7 papers with code

Cost-Effective Retraining of Machine Learning Models

no code implementations6 Oct 2023 Ananth Mahadevan, Michael Mathioudakis

Our main contribution is a Cost-Aware Retraining Algorithm called Cara, which optimizes the trade-off over streams of data and queries.

WaZI: A Learned and Workload-aware Z-Index

no code implementations6 Oct 2023 Sachith Pai, Michael Mathioudakis, Yanhao Wang

Specifically, we first formulate a cost function to measure the performance of a Z-index on a dataset for a range-query workload.

Graph Summarization via Node Grouping: A Spectral Algorithm

1 code implementation8 Nov 2022 Arpit Merchant, Michael Mathioudakis, Yanhao Wang

By initially allowing relaxed (fractional) solutions for integer maximization, we analytically expose the underlying connections to the spectral properties of the adjacency matrix.

Streaming Algorithms for Diversity Maximization with Fairness Constraints

1 code implementation30 Jul 2022 Yanhao Wang, Francesco Fabbri, Michael Mathioudakis

Given a set $X$ of $n$ elements, it asks to select a subset $S$ of $k \ll n$ elements with maximum \emph{diversity}, as quantified by the dissimilarities among the elements in $S$.

Attribute Data Summarization +2

Rewiring What-to-Watch-Next Recommendations to Reduce Radicalization Pathways

1 code implementation1 Feb 2022 Francesco Fabbri, Yanhao Wang, Francesco Bonchi, Carlos Castillo, Michael Mathioudakis

Hence, we define the problem of reducing the prevalence of radicalization pathways by selecting a small number of edges to "rewire", so to minimize the maximum of segregation scores among all radicalized nodes, while maintaining the relevance of the recommendations.

Recommendation Systems

Certifiable Machine Unlearning for Linear Models

1 code implementation29 Jun 2021 Ananth Mahadevan, Michael Mathioudakis

In this paper, we present an experimental study of the three state-of-the-art approximate unlearning methods for linear models and demonstrate the trade-offs between efficiency, effectiveness and certifiability offered by each method.

Machine Unlearning

Joint Use of Node Attributes and Proximity for Semi-Supervised Classification on Graphs

no code implementations22 Oct 2020 Arpit Merchant, Michael Mathioudakis

The task of node classification is to infer unknown node labels, given the labels for some of the nodes along with the network structure and other node attributes.

General Classification Node Classification

Fair and Representative Subset Selection from Data Streams

1 code implementation9 Oct 2020 Yanhao Wang, Francesco Fabbri, Michael Mathioudakis

We study the problem of extracting a small subset of representative items from a large data stream.

Data Summarization Fairness +1

GRMR: Generalized Regret-Minimizing Representatives

no code implementations19 Jul 2020 Yanhao Wang, Michael Mathioudakis, Yuchen Li, Kian-Lee Tan

Extracting a small subset of representative tuples from a large database is an important task in multi-criteria decision making.

Data Structures and Algorithms Databases

Towards Data-Driven Affirmative Action Policies under Uncertainty

no code implementations2 Jul 2020 Corinna Hertweck, Carlos Castillo, Michael Mathioudakis

In this paper, we study university admissions under a centralized system that uses grades and standardized test scores to match applicants to university programs.

Political Discourse on Social Media: Echo Chambers, Gatekeepers, and the Price of Bipartisanship

1 code implementation5 Jan 2018 Kiran Garimella, Gianmarco De Francisci Morales, Aristides Gionis, Michael Mathioudakis

By comparing the two, we find that Twitter users are, to a large degree, exposed to political opinions that agree with their own.

Social and Information Networks

Quantifying Controversy in Social Media

1 code implementation18 Jul 2015 Kiran Garimella, Gianmarco De Francisci Morales, Aristides Gionis, Michael Mathioudakis

Unlike previous work, rather than study controversy in a single hand-picked topic and use domain specific knowledge, we take a general approach to study topics in any domain.

Social and Information Networks

Cannot find the paper you are looking for? You can Submit a new open access paper.