no code implementations • 23 Feb 2024 • Abolfazl Younesi, Mohsen Ansari, Mohammadamin Fazli, Alireza Ejlali, Muhammad Shafique, Jörg Henkel
In today's digital age, Convolutional Neural Networks (CNNs), a subset of Deep Learning (DL), are widely used for various computer vision tasks such as image classification, object detection, and image segmentation.
no code implementations • 12 Feb 2024 • Dimitrios Danopoulos, Georgios Zervakis, Dimitrios Soudris, Jörg Henkel
Vision Transformer (ViT) models which were recently introduced by the transformer architecture have shown to be very competitive and often become a popular alternative to Convolutional Neural Networks (CNNs).
no code implementations • 23 Dec 2023 • Konstantinos Balaskas, Andreas Karatzas, Christos Sad, Kostas Siozios, Iraklis Anagnostopoulos, Georgios Zervakis, Jörg Henkel
We explore, for the first time, per-layer fine- and coarse-grained pruning, in the same DNN architecture, in addition to low bit-width mixed-precision quantization for weights and activations.
no code implementations • 18 Jul 2023 • Kilian Pfeiffer, Martin Rapp, Ramin Khalili, Jörg Henkel
With an increasing number of smart devices like internet of things (IoT) devices deployed in the field, offloadingtraining of neural networks (NNs) to a central server becomes more and more infeasible.
1 code implementation • NeurIPS 2023 • Kilian Pfeiffer, Ramin Khalili, Jörg Henkel
If the required memory to train a model exceeds this limit, the device will be excluded from the training.
1 code implementation • 14 Mar 2023 • Giorgos Armeniakos, Georgios Zervakis, Dimitrios Soudris, Mehdi B. Tahoori, Jörg Henkel
Printed electronics (PE) promises on-demand fabrication, low non-recurring engineering costs, and sub-cent fabrication costs.
1 code implementation • 28 Feb 2023 • Giorgos Armeniakos, Georgios Zervakis, Dimitrios Soudris, Mehdi B. Tahoori, Jörg Henkel
Printed Electronics (PE) exhibits on-demand, extremely low-cost hardware due to its additive manufacturing process, enabling machine learning (ML) applications for domains that feature ultra-low cost, conformity, and non-toxicity requirements that silicon-based systems cannot deliver.
no code implementations • 2 Oct 2022 • Jörg Henkel, Hai Li, Anand Raghunathan, Mehdi B. Tahoori, Swagath Venkataramani, Xiaoxuan Yang, Georgios Zervakis
In this work, we enlighten the synergistic nature of AxC and ML and elucidate the impact of AxC in designing efficient ML systems.
no code implementations • 25 Jul 2022 • Ourania Spantidi, Georgios Zervakis, Iraklis Anagnostopoulos, Jörg Henkel
Deep Neural Networks (DNNs) are being heavily utilized in modern applications and are putting energy-constraint devices to the test.
no code implementations • 16 Mar 2022 • Giorgos Armeniakos, Georgios Zervakis, Dimitrios Soudris, Jörg Henkel
To enable efficient execution of DNN inference, more and more research works, therefore, exploit the inherent error resilience of DNNs and employ Approximate Computing (AC) principles to address the elevated energy demands of DNN accelerators.
1 code implementation • 11 Mar 2022 • Giorgos Armeniakos, Georgios Zervakis, Dimitrios Soudris, Mehdi B. Tahoori, Jörg Henkel
In our extensive experimental evaluation we consider 14 MLPs and SVMs and evaluate more than 4300 approximate and exact designs.
1 code implementation • 10 Mar 2022 • Kilian Pfeiffer, Martin Rapp, Ramin Khalili, Jörg Henkel
To adapt to the devices' heterogeneous resources, CoCoFL freezes and quantizes selected layers, reducing communication, computation, and memory requirements, whereas other layers are still trained in full precision, enabling to reach a high accuracy.
1 code implementation • 8 Mar 2022 • Dimitrios Danopoulos, Georgios Zervakis, Kostas Siozios, Dimitrios Soudris, Jörg Henkel
Current state-of-the-art employs approximate multipliers to address the highly increased power demands of DNN accelerators.
no code implementations • 16 Dec 2021 • Martin Rapp, Ramin Khalili, Kilian Pfeiffer, Jörg Henkel
We study the problem of distributed training of neural networks (NNs) on devices with heterogeneous, limited, and time-varying availability of computational resources.
no code implementations • 20 Jul 2021 • Ourania Spantidi, Georgios Zervakis, Iraklis Anagnostopoulos, Hussam Amrouch, Jörg Henkel
In addition, we propose a filter-oriented approximation method to map the weights to the appropriate modes of the approximate multiplier.
no code implementations • 8 Mar 2021 • Sami Salamin, Georgios Zervakis, Ourania Spantidi, Iraklis Anagnostopoulos, Jörg Henkel, Hussam Amrouch
Transistor aging is one of the major concerns that challenges designers in advanced technologies.
no code implementations • 18 Feb 2021 • Georgios Zervakis, Ourania Spantidi, Iraklis Anagnostopoulos, Hussam Amrouch, Jörg Henkel
In this work, we introduce a control variate approximation technique for low error approximate Deep Neural Network (DNN) accelerators.
no code implementations • 9 Jun 2020 • Martin Rapp, Ramin Khalili, Jörg Henkel
We consider a distributed system, consisting of a heterogeneous set of devices, ranging from low-end to high-end.
no code implementations • 6 Apr 2020 • Virinchi Roy Surabhi, Prashanth Krishnamurthy, Hussam Amrouch, Kanad Basu, Jörg Henkel, Ramesh Karri, Farshad Khorrami
Combining IC aging with over-clocking produces a pattern of bit errors at the IC output by the induced timing violations.