Inference of Large Language Models (LLMs) across computer clusters has become a focal point of research in recent times, with many acceleration techniques taking inspiration from CPU speculative execution. These techniques reduce bottlenecks associated ...

Today’s performance analysis frameworks for deep learning accelerators suffer from two significant limitations. First, although modern convolutional neural networks (CNNs) consist of many types of layers other than convolution, especially during training, ...

There are many issues that can cause problems when attempting to infer model parameters from data. Data and models are both imperfect, and as such there are multiple scenarios in which standard methods of inference will lead to misleading conclusions; ...

Partitioning Artificial Intelligence (AI) models such as Deep Neural Networks (DNNs) or Transformer-based Architectures is essential for minimizing latency in resource-constrained edge computing environments, which is critical for applications such as ...

This paper presents a case study focused on synthesizing relational datasets within Official Statistics for software and technology testing purposes. Specifically, the focus is on generating synthetic data for testing and validating software code. ...

We delve into the performance of transformer encoder inference on low-power multi-core processors from two perspectives: First, we conduct a detailed profile of the inference process for two members of the BERT family on a modern multi-core ...

DNN models are becoming increasingly large and complex, but they are also being deployed on commodity devices that require low power and latency but lack specialized accelerators. We introduce LUTIN (LUT-based INference), which reduces the amount of ...

Navigating the intricacies of data management in the era of Large Language Models (LLMs) presents both challenges and opportunities for database and data management communities. In this tutorial, we offer a comprehensive exploration into the vital role ...

Animation as genre is broadly used across many forms of digital media. In this paper, I argue ChatGPT and similar chatbots powered by Large Language Models (LLMs) can be best understood as animated characters. More than just cartooning, puppetry, or CGI, ...

Personalization improves user experience by tailoring interactions relevant to each user’s background and preferences. However, personalization requires information about users that platforms often collect without their awareness or their enthusiastic ...

In this paper, we present a comprehensive empirical study to evaluate four prominent Computer Vision inference frameworks. Our goal is to shed light on their strengths and weaknesses and provide valuable insights into the challenges of selecting the ...

Co-locating multiple DNN servings to share GPU resource is widely used to improve resource utilization while guaranteeing user QoS. Existing GPU sharing mechanism is restricted to model level, and fluctuations in kernel-level resource demands highlight a ...

Transformer is a popular machine learning model used by many intelligent applications in smart cities. However, it has high computational complexity and it would be hard to deploy it in weak-edge devices. This paper presents a novel two-round offloading ...

Computing the posterior distribution of a probabilistic program is a hard task for which no one-fit-for-all solution exists. We propose Gaussian Semantics, which approximates the exact probabilistic semantics of a bounded program by means of Gaussian ...

This paper introduces a novel framework that eliminates the often cumbersome "build and install" step when running software. Our framework packages a collection of techniques to automatically infer and generate sandboxes, specifically Linux ...

I propose a simple forward difference-in-differences method to estimate causal effects from quasiexperimental data and also develop its inference theory, which is widely applicable.

The difference-in-differences (DID) method is the most widely used tool to answer causal questions from quasiexperimental data in marketing and the broader social sciences. Because assignment to treatment in quasiexperiments is not random, the selection ...

‘‘Extreme edge”¹ devices, such as smart sensors, are a uniquely challenging environment for the deployment of machine learning. The tiny energy budgets of these devices lie beyond what is feasible for conventional deep neural networks, particularly in ...

Serverless offers a scalable and cost-effective service model for users to run applications without focusing on underlying infrastructure or physical servers. While the Serverless architecture is not designed to address the unique challenges posed by ...

Applying services computing to neural networks, a service provider may provide inference with a pre-trained neural network as a service. Clients use the service to get the neural network's output on their input. To protect sensitive data, secure neural ...

NullPointerExceptions (NPEs), caused by dereferencing null, fre- quently cause crashes in Java programs. Pluggable type checking is highly effective in preventing Java NPEs. However, this approach is difficult to adopt for large, existing code bases, as ...