-
Towards Neural Architecture Search for Transfer Learning in 6G Networks
Authors:
Adam Orucu,
Farnaz Moradi,
Masoumeh Ebrahimi,
Andreas Johnsson
Abstract:
The future 6G network is envisioned to be AI-native, and as such, ML models will be pervasive in support of optimizing performance, reducing energy consumption, and in co** with increasing complexity and heterogeneity. A key challenge is automating the process of finding optimal model architectures satisfying stringent requirements stemming from varying tasks, dynamicity and available resources…
▽ More
The future 6G network is envisioned to be AI-native, and as such, ML models will be pervasive in support of optimizing performance, reducing energy consumption, and in co** with increasing complexity and heterogeneity. A key challenge is automating the process of finding optimal model architectures satisfying stringent requirements stemming from varying tasks, dynamicity and available resources in the infrastructure and deployment positions. In this paper, we describe and review the state-of-the-art in Neural Architecture Search and Transfer Learning and their applicability in networking. Further, we identify open research challenges and set directions with a specific focus on three main requirements with elements unique to the future network, namely combining NAS and TL, multi-objective search, and tabular data. Finally, we outline and discuss both near-term and long-term work ahead.
△ Less
Submitted 4 June, 2024;
originally announced June 2024.
-
Using Large Language Models to Understand Telecom Standards
Authors:
Athanasios Karapantelakis,
Mukesh Thakur,
Alexandros Nikou,
Farnaz Moradi,
Christian Orlog,
Fitsum Gaim,
Henrik Holm,
Doumitrou Daniil Nimara,
Vincent Huang
Abstract:
The Third Generation Partnership Project (3GPP) has successfully introduced standards for global mobility. However, the volume and complexity of these standards has increased over time, thus complicating access to relevant information for vendors and service providers. Use of Generative Artificial Intelligence (AI) and in particular Large Language Models (LLMs), may provide faster access to releva…
▽ More
The Third Generation Partnership Project (3GPP) has successfully introduced standards for global mobility. However, the volume and complexity of these standards has increased over time, thus complicating access to relevant information for vendors and service providers. Use of Generative Artificial Intelligence (AI) and in particular Large Language Models (LLMs), may provide faster access to relevant information. In this paper, we evaluate the capability of state-of-art LLMs to be used as Question Answering (QA) assistants for 3GPP document reference. Our contribution is threefold. First, we provide a benchmark and measuring methods for evaluating performance of LLMs. Second, we do data preprocessing and fine-tuning for one of these LLMs and provide guidelines to increase accuracy of the responses that apply to all LLMs. Third, we provide a model of our own, TeleRoBERTa, that performs on-par with foundation LLMs but with an order of magnitude less number of parameters. Results show that LLMs can be used as a credible reference tool on telecom technical documents, and thus have potential for a number of different applications from troubleshooting and maintenance, to network operations and software product development.
△ Less
Submitted 12 April, 2024; v1 submitted 2 April, 2024;
originally announced April 2024.
-
Spin-Orbit Torque Flash Analog-to-Digital Converter
Authors:
Hamdam Ghanatian,
Luana Benetti,
Pedro Anacleto,
Tim Bohnert,
Hooman Farkhani,
Ricardo Ferreira,
Farshad Moradi
Abstract:
Although Analog-to-digital converters (ADCs) are critical components in mixed-signal integrated circuits (IC), their performance has not been improved significantly over the last decade. To achieve a radical improvement (compact, low power and reliable ADCs), spintronics can be considered as a proper candidate due to its compatibility with CMOS and wide applications in storage, neuromorphic comput…
▽ More
Although Analog-to-digital converters (ADCs) are critical components in mixed-signal integrated circuits (IC), their performance has not been improved significantly over the last decade. To achieve a radical improvement (compact, low power and reliable ADCs), spintronics can be considered as a proper candidate due to its compatibility with CMOS and wide applications in storage, neuromorphic computing, and so on. In this paper, a proof-of-concept of a 3-bit spin-CMOS Flash ADC using in-plane-anisotropy magnetic tunnel junctions (i-MTJs) with spin-orbit torque (SOT) switching mechanism is designed, fabricated and characterized. The proposed ADC replaces the current mirrors and power-hungry comparators in the conventional Flash ADC with seven parallel i-MTJs with different heavy metal (HM) widths. Monte-Carlo simulations based on the experimental measurements show the process variations/mismatch limits the accuracy of the proposed ADC to 2 bits. Moreover, the maximum differential nonlinearity (DNL) and integral nonlinearity (INL) are 0.739 LSB (least significant bit) and 0.7319 LSB, respectively.
△ Less
Submitted 9 January, 2023;
originally announced January 2023.
-
NET-TEN: a silicon neuromorphic network for low-latency detection of seizures in local field potentials
Authors:
Margherita Ronchini,
Yasser Rezaeiyan,
Milad Zamani,
Gabriella Panuccio,
Farshad Moradi
Abstract:
Therapeutic intervention in neurological disorders still relies heavily on pharmacological solutions, while the treatment of patients with drug resistance remains an open challenge. This is particularly true for patients with epilepsy, 30% of whom are refractory to medications. Implantable devices for chronic recording and electrical modulation of brain activity have proved a viable alternative in…
▽ More
Therapeutic intervention in neurological disorders still relies heavily on pharmacological solutions, while the treatment of patients with drug resistance remains an open challenge. This is particularly true for patients with epilepsy, 30% of whom are refractory to medications. Implantable devices for chronic recording and electrical modulation of brain activity have proved a viable alternative in such cases. To operate, the device should detect the relevant electrographic biomarkers from Local Field Potentials (LFPs) and determine the right time for stimulation. To enable timely interventions, the ideal device should attain biomarker detection with low latency while operating under low power consumption to prolong the battery life. Neuromorphic networks have progressively gained reputation as low-latency low-power computing systems, which makes them a promising candidate as processing core of next-generation implantable neural interfaces. Here we introduce a fully-analog neuromorphic device implemented in CMOS technology for analyzing LFP signals in an in vitro model of acute ictogenesis. We show that the system can detect ictal and interictal events with ms-latency and with high precision, consuming on average 3.50 nW during the task. Our work paves the way to a new generation of brain implantable devices for personalized closed-loop stimulation for epilepsy treatment.
△ Less
Submitted 19 October, 2022;
originally announced October 2022.
-
In the realm of hybrid Brain: Human Brain and AI
Authors:
Hoda Fares,
Margherita Ronchini,
Milad Zamani,
Hooman Farkhani,
Farshad Moradi
Abstract:
With the recent developments in neuroscience and engineering, it is now possible to record brain signals and decode them. Also, a growing number of stimulation methods have emerged to modulate and influence brain activity. Current brain-computer interface (BCI) technology is mainly on therapeutic outcomes, it already demonstrated its efficiency as assistive and rehabilitative technology for patien…
▽ More
With the recent developments in neuroscience and engineering, it is now possible to record brain signals and decode them. Also, a growing number of stimulation methods have emerged to modulate and influence brain activity. Current brain-computer interface (BCI) technology is mainly on therapeutic outcomes, it already demonstrated its efficiency as assistive and rehabilitative technology for patients with severe motor impairments. Recently, artificial intelligence (AI) and machine learning (ML) technologies have been used to decode brain signals. Beyond this progress, combining AI with advanced BCIs in the form of implantable neurotechnologies grants new possibilities for the diagnosis, prediction, and treatment of neurological and psychiatric disorders. In this context, we envision the development of closed loop, intelligent, low-power, and miniaturized neural interfaces that will use brain inspired AI techniques with neuromorphic hardware to process the data from the brain. This will be referred to as Brain Inspired Brain Computer Interfaces (BI-BCIs). Such neural interfaces would offer access to deeper brain regions and better understanding for brain's functions and working mechanism, which improves BCIs operative stability and system's efficiency. On one hand, brain inspired AI algorithms represented by spiking neural networks (SNNs) would be used to interpret the multimodal neural signals in the BCI system. On the other hand, due to the ability of SNNs to capture rich dynamics of biological neurons and to represent and integrate different information dimensions such as time, frequency, and phase, it would be used to model and encode complex information processing in the brain and to provide feedback to the users. This paper provides an overview of the different methods to interface with the brain, presents future applications and discusses the merger of AI and BCIs.
△ Less
Submitted 17 January, 2024; v1 submitted 4 October, 2022;
originally announced October 2022.
-
Towards Real-World BCI: CCSPNet, A Compact Subject-Independent Motor Imagery Framework
Authors:
Mahbod Nouri,
Faraz Moradi,
Hafez Ghaemi,
Ali Motie Nasrabadi
Abstract:
A conventional brain-computer interface (BCI) requires a complete data gathering, training, and calibration phase for each user before it can be used. In recent years, a number of subject-independent (SI) BCIs have been developed. Many of these methods yield a weaker performance compared to the subject-dependent (SD) approach, and some are computationally expensive. A potential real-world applicat…
▽ More
A conventional brain-computer interface (BCI) requires a complete data gathering, training, and calibration phase for each user before it can be used. In recent years, a number of subject-independent (SI) BCIs have been developed. Many of these methods yield a weaker performance compared to the subject-dependent (SD) approach, and some are computationally expensive. A potential real-world application would greatly benefit from a more accurate, compact, and computationally efficient subject-independent BCI. In this work, we propose a novel subject-independent BCI framework, named CCSPNet (Convolutional Common Spatial Pattern Network) that is trained on the motor imagery (MI) paradigm of a large-scale electroencephalography (EEG) signals database consisting of 400 trials for every 54 subjects who perform two-class hand-movement MI tasks. The proposed framework applies a wavelet kernel convolutional neural network (WKCNN) and a temporal convolutional neural network (TCNN) in order to represent and extract the spectral features of EEG signals. A common spatial pattern (CSP) algorithm is implemented for spatial feature extraction, and the number of CSP features is reduced by a dense neural network. Finally, the class label is determined by a linear discriminant analysis (LDA) classifier. The CCSPNet evaluation results show that it is possible to have a compact BCI that achieves both SD and SI state-of-the-art performance comparable to complex and computationally expensive models.
△ Less
Submitted 8 October, 2022; v1 submitted 25 December, 2020;
originally announced December 2020.
-
Spin-Orbit-Torque-based Devices, Circuits and Architectures
Authors:
Farshad Moradi,
Hooman Farkhani,
Behzad Zeinali,
Hamdam Ghanatian,
Johan Michel Alain Pelloux-Prayer,
Tim Boehnert,
Mohammad Zahedinejad,
Hadi Heidari,
Vahid Nabaei,
Ricardo Ferreira,
Johan Akerman,
Jens Kargaard Madsen
Abstract:
Spintronics, the use of spin of an electron instead of its charge, has received huge attention from research communities for different applications including memory, interconnects, logic implementation, neuromorphic computing, and many other applications. Here, in this paper, we review the works within spintronics, more specifically on spin-orbit torque (SOT) within different research groups. We a…
▽ More
Spintronics, the use of spin of an electron instead of its charge, has received huge attention from research communities for different applications including memory, interconnects, logic implementation, neuromorphic computing, and many other applications. Here, in this paper, we review the works within spintronics, more specifically on spin-orbit torque (SOT) within different research groups. We also provide researchers an insight into the future potentials of the SOT-based designs. This comprehensive review paper covers different aspects of SOT-based design from device and circuit to architecture level as well as more ambitious and futuristic applications of such technology.
△ Less
Submitted 3 December, 2019;
originally announced December 2019.
-
Analyzing the Social Structure and Dynamics of E-mail and Spam in Massive Backbone Internet Traffic
Authors:
Farnaz Moradi,
Tomas Olovsson,
Philippas Tsigas
Abstract:
E-mail is probably the most popular application on the Internet, with everyday business and personal communications dependent on it. Spam or unsolicited e-mail has been estimated to cost businesses significant amounts of money. However, our understanding of the network-level behavior of legitimate e-mail traffic and how it differs from spam traffic is limited. In this study, we have passively capt…
▽ More
E-mail is probably the most popular application on the Internet, with everyday business and personal communications dependent on it. Spam or unsolicited e-mail has been estimated to cost businesses significant amounts of money. However, our understanding of the network-level behavior of legitimate e-mail traffic and how it differs from spam traffic is limited. In this study, we have passively captured SMTP packets from a 10 Gbit/s Internet backbone link to construct a social network of e-mail users based on their exchanged e-mails. The focus of this paper is on the graph metrics indicating various structural properties of e-mail networks and how they evolve over time. This study also looks into the differences in the structural and temporal characteristics of spam and non-spam networks. Our analysis on the collected data allows us to show several differences between the behavior of spam and legitimate e-mail traffic, which can help us to understand the behavior of spammers and give us the knowledge to statistically model spam traffic on the network-level in order to complement current spam detection techniques.
△ Less
Submitted 19 August, 2010;
originally announced August 2010.
