no code implementations • 1 Dec 2024 • Mahdi Erfanian, Mohsen Dehghankar, Abolfazl Asudeh
Following this method, we develop a system for image data retrieval and propose practical solutions that enable leveraging future advancements in GenAI and vector representations for improving our system's performance.
no code implementations • 30 Nov 2024 • Mohsen Dehghankar, Abolfazl Asudeh
Observing that LLMs are less likely to miss elements at certain positions of the input, we introduce the problem of LLM input reranking: to find a ranking of the input that maximizes the LLM's accuracy for the given query without making explicit assumptions about the query.
1 code implementation • 10 Nov 2024 • Mohsen Dehghankar, Mahdi Erfanian, Abolfazl Asudeh
To address these challenges and make LLMs more accessible and cost-effective, in this paper, we propose algorithms to improve the inference time and memory efficiency of 1. 58-bit LLMs with ternary weight matrices.
1 code implementation • 7 Nov 2024 • Mohsen Dehghankar, Abolfazl Asudeh
Specifically, we propose a minority mining problem, where we find vectors in the attribute space that reveal potential groups that are under-represented and under-performing.
no code implementations • 17 Apr 2024 • Sana Ebrahimi, Nima Shahbazi, Abolfazl Asudeh
Towards addressing this gap, we introduce REQUAL-LM, a novel method for finding reliable and equitable LLM outputs through aggregation.
1 code implementation • 10 Apr 2024 • Nima Shahbazi, Mahdi Erfanian, Abolfazl Asudeh, Fatemeh Nargesian, Divesh Srivastava
Entity matching is one the earliest tasks that occur in the big data pipeline and is alarmingly exposed to unintentional biases that affect the quality of data.
no code implementations • 1 Mar 2024 • Sana Ebrahimi, Kaiwen Chen, Abolfazl Asudeh, Gautam Das, Nick Koudas
Pre-trained Large Language Models (LLMs) have significantly advanced natural language processing capabilities but are susceptible to biases present in their training data, leading to unfair outcomes in various applications.
1 code implementation • 2 Feb 2024 • Mahdi Erfanian, H. V. Jagadish, Abolfazl Asudeh
The potential harms of the under-representation of minorities in training data, particularly in multi-modal settings, is a well-recognized concern.
no code implementations • 14 Jul 2023 • Suraj Shetiya, Shohedul Hasan, Abolfazl Asudeh, Gautam Das
Linear Regression is a seminal technique in statistics and machine learning, where the objective is to build linear predictive models between a response (i. e., dependent) variable and one or more predictor (i. e., independent) variables.
1 code implementation • 6 Jul 2023 • Nima Shahbazi, Nikola Danevski, Fatemeh Nargesian, Abolfazl Asudeh, Divesh Srivastava
Entity matching (EM) is a challenging problem studied by different communities for over half a century.
no code implementations • 15 Jun 2023 • Sana Ebrahimi, Rishi Advani, Abolfazl Asudeh
Making a connection between the two research directions as special cases of approximating matrix multiplications in the context of neural networks, we provide a negative theoretical analysis that shows feedforward approximation is an obstacle against scalability.
no code implementations • 10 Apr 2023 • Francesco Di Carlo, Nazanin Nezami, Hadis Anahideh, Abolfazl Asudeh
Despite the potential benefits of machine learning (ML) in high-risk decision-making domains, the deployment of ML is not accessible to practitioners, and there is a risk of discrimination.
no code implementations • 22 Mar 2022 • Nima Shahbazi, Yin Lin, Abolfazl Asudeh, H. V. Jagadish
Data-driven algorithms are only as good as the data they work with, while data sets, especially social data, often fail to represent minorities adequately.
no code implementations • 13 Sep 2021 • Hadis Anahideh, Nazanin Nezami, Abolfazl Asudeh
Using the estimated correlations, we then find a subset of representative metrics.
no code implementations • 13 Mar 2021 • Hantian Zhang, Xu Chu, Abolfazl Asudeh, Shamkant B. Navathe
Existing techniques for producing fair ML models either are limited to the type of fairness constraints they can handle (e. g., preprocessing) or require nontrivial modifications to downstream ML training algorithms (e. g., in-processing).
1 code implementation • 20 Jun 2020 • Hadis Anahideh, Abolfazl Asudeh, Saravanan Thirumuruganathan
Collecting accurate labeled data in societal applications is challenging and costly.
1 code implementation • 6 Jan 2020 • Hadis Anahideh, Abolfazl Asudeh, Saravanan Thirumuruganathan
Machine learning (ML) is increasingly being used in high-stakes applications impacting society.
no code implementations • 22 Nov 2019 • Abolfazl Asudeh, H. V. Jagadish
We provide unbiased samplers for the entire function space, as well as a $\theta$-vicinity around a given function.
no code implementations • 15 Sep 2014 • Abolfazl Asudeh, Gensheng Zhang, Naeemul Hassan, Chengkai Li, Gergely V. Zaruba
This design is both sufficient and efficient, as it is proven to find a short terminal question sequence.