Related papers: Towards AI-Native Fronthaul: Neural Compression fo…

Recent advances in deep learning have made available large, powerful convolutional neural networks (CNN) with state-of-the-art performance in several real-world applications. Unfortunately, these large-sized models have millions of…

This paper presents a new deep learning architecture for Natural Language Inference (NLI). Firstly, we introduce a new architecture where alignment pairs are compared, compressed and then propagated to upper layers for enhanced…

Compute-in-memory (CIM) based neural network accelerators offer a promising solution to the Von Neumann bottleneck by computing directly within memory arrays. However, SRAM CIM faces limitations in executing larger models due to its cell…

This paper aims to characterize the synergy of distributed caching and wireless fronthaul in a fog radio access network (Fog-RAN) where all edge nodes (ENs) and user equipments (UEs) have a local cache and store contents independently at…

In this paper, a new queuing model for the Tactile Internet (TI) is proposed for the cloud radio access network (C-RAN) architecture of the next generation wireless networks, e.g., 5G, assisted via orthogonal frequency division multiple…

Cloud Radio Access Network (C-RAN) refers to the virtualization of base station functionalities by means of cloud computing. This results in a novel cellular architecture in which low-cost wireless access points, known as radio units (RUs)…

In this work, we propose a simple but effective channel pruning framework called Progressive Channel Pruning (PCP) to accelerate Convolutional Neural Networks (CNNs). In contrast to the existing channel pruning methods that prune channels…

The fronthaul connection is a key component of Centralized RAN (C-RAN) architectures, consistently required to handle high capacity demands. However, this critical feature is at risk when the transport link relies on wireless technology.…

Model compression aims to reduce the redundancy of deep networks to obtain compact models. Recently, channel pruning has become one of the predominant compression methods to deploy deep models on resource-constrained devices. Most channel…

Deploying trained convolutional neural networks (CNNs) to mobile devices is a challenging task because of the simultaneous requirements of the deployed model to be fast, lightweight and accurate. Designing and training a CNN architecture…

Artificial intelligence (AI)-native radio access networks (RANs) will serve vertical industries with stringent requirements: smart grids, autonomous vehicles, remote healthcare, industrial automation, etc. To achieve these requirements,…

While Convolutional Neural Networks (CNNs) excel at learning complex latent-space representations, their over-parameterization can lead to overfitting and reduced performance, particularly with limited data. This, alongside their high…

This dissertation paper presents the main contributions to the design and the implementation of a Cloud-RAN solution. We concretely address the two main challenges of Cloud-RAN systems: real-time processing of radio signals and reduced…

Large Deep Neural Networks (DNNs) are the backbone of today's artificial intelligence due to their ability to make accurate predictions when being trained on huge datasets. With advancing technologies, such as the Internet of Things,…

Many common types of data can be represented as functions that map coordinates to signal values, such as pixel locations to RGB values in the case of an image. Based on this view, data can be compressed by overfitting a compact neural…

In this paper, we propose an energy-efficient federated learning (FL) framework for the energy-constrained devices over cloud radio access network (Cloud-RAN), where each device adopts quantized neural networks (QNNs) to train a local FL…

Mobile and embedded machine learning developers frequently have to compromise between two inferior on-device deployment strategies: sacrifice accuracy and aggressively shrink their models to run on dedicated low-power cores; or sacrifice…

Deep Neural Networks (DNNs) have shown significant advantages in a wide variety of domains. However, DNNs are becoming computationally intensive and energy hungry at an exponential pace, while at the same time, there is a vast demand for…

As the industry deploys increasingly large and complex neural networks to mobile devices, more pressure is put on the memory and compute resources of those devices. Deep compression, or compression of deep neural network weight matrices, is…

Modern deep learning models are often trained in parallel over a collection of distributed machines to reduce training time. In such settings, communication of model updates among machines becomes a significant performance bottleneck and…