Despite their tremendous success and versatility, Deep Neural Networks (DNNs) such as Large Language Models (LLMs) suffer from inference inefficiency and rely on advanced computational infrastructure. To address these challenges and make these models more accessible and cost-effective, in this paper, we propose algorithms to improve the inference time and memory efficiency of DNNs with binary and ternary weight matrices. Particularly focusing on matrix multiplication as the bottleneck operation of inference, we observe that, once trained, the weight matrices of a model no longer change. This allows us to preprocess these matrices and create indices that help reduce the storage requirements by a logarithmic factor while enabling our efficient inference algorithms. Specifically, for a $n\times n$ weight matrix, our efficient algorithm guarantees a time complexity of $O(\frac{n^2}{\log n})$, a logarithmic factor improvement over the standard vector-matrix multiplication. Besides theoretical analysis, we conduct extensive experiments to evaluate the practical efficiency of our algorithms. Our results confirm the superiority of our approach both with respect to time and memory, as we observed a reduction in the multiplication time up to 29x and memory usage up to 6x. When applied to LLMs, our experiments show up to a 5.24x speedup in the inference time.

Potential harms from the under-representation of minorities in data, particularly in multi-modal settings, is a well-recognized concern. While there has been extensive effort in detecting such under-representation, resolution has remained a challenge. With recent generative AI advancements, large language and foundation models have emerged as versatile tools across various domains. In this paper, we propose Chameleon, a system that efficiently utilizes these tools to augment a dataset with minimal addition of synthetically generated tuples to enhance the coverage of the under-represented groups. Our system applies quality and outlier-detection tests to ensure the quality and semantic integrity of the generated tuples. In order to minimize the rejection chance of the generated tuples, we propose multiple strategies to provide a guide for the foundation model. Our experiment results, in addition to confirming the efficiency of our proposed algorithms, illustrate our approach's effectiveness, as the model's unfairness in a downstream task significantly dropped after data repair using Chameleon.

Entity matching is one of the earliest tasks that occur in the big data pipeline and is alarmingly exposed to unintentional biases that affect the quality of data. Identifying and mitigating the biases that exist in the data or are introduced by the matcher at this stage can contribute to promoting fairness in downstream tasks. This demonstration showcases FairEM360, a framework for 1) auditing the output of entity matchers across a wide range of fairness measures and paradigms, 2) providing potential explanations for the underlying reasons for unfairness, and 3) providing resolutions for the unfairness issues through an exploratory process with human-in-the-loop feedback, utilizing an ensemble of matchers. We aspire for FairEM360 to contribute to the prioritization of fairness as a key consideration in the evaluation of EM pipelines.

Graph Pattern Mining (GPM) refers to a class of problems involving the processing of sub graphs extracted from larger graphs. Applications to GPM algorithms include querying subgraphs, identifying motif structures in biological networks, characterizing social media, among others. G PM algorithms are challenging to develop due to subroutines that include non-trivial graph theory concepts and methods such as isomorphism. General-purpose GPM systems have emerged as a solution to improve the user experience with such algorithms. However, existing general-purpose GPM systems are heterogeneous in terms of implementation details, hardware environment and algorithmic paradigms for sub graph exploration and thus, observations taken from the experimental results alone may not clearly identify when a particular paradigm prevails over another. In this work we present an experimentation analysis of popular paradigms used in existing GPM systems. In order to provide a fair and comprehensive evaluation of various algorithmic paradigms we implement all of them within a single GPM framework. Our results show that no single paradigm is best for every application scenario, and we believe that our findings may guide practitioner towards more optimized GPM systems in the future.