In-network machine learning using programmable network devices: A survey
Machine learning is widely used to solve networking challenges, ranging from traffic
classification and anomaly detection to network configuration. However, machine learning …
classification and anomaly detection to network configuration. However, machine learning …
Hardware-assisted machine learning in resource-constrained IoT environments for security: review and future prospective
G Kornaros - IEEE Access, 2022 - ieeexplore.ieee.org
As the Internet of Things (IoT) technology advances, billions of multidisciplinary smart
devices act in concert, rarely requiring human intervention, posing significant challenges in …
devices act in concert, rarely requiring human intervention, posing significant challenges in …
8-bit optimizers via block-wise quantization
Stateful optimizers maintain gradient statistics over time, eg, the exponentially smoothed
sum (SGD with momentum) or squared sum (Adam) of past gradient values. This state can …
sum (SGD with momentum) or squared sum (Adam) of past gradient values. This state can …
Scaling distributed machine learning with {In-Network} aggregation
Training machine learning models in parallel is an increasingly important workload. We
accelerate distributed parallel training by designing a communication primitive that uses a …
accelerate distributed parallel training by designing a communication primitive that uses a …
Floatpim: In-memory acceleration of deep neural network training with high precision
Processing In-Memory (PIM) has shown a great potential to accelerate inference tasks of
Convolutional Neural Network (CNN). However, existing PIM architectures do not support …
Convolutional Neural Network (CNN). However, existing PIM architectures do not support …
Communication compression for decentralized training
Optimizing distributed learning systems is an art of balancing between computation and
communication. There have been two lines of research that try to deal with slower …
communication. There have been two lines of research that try to deal with slower …
Pushing the limits of narrow precision inferencing at cloud scale with microsoft floating point
B Darvish Rouhani, D Lo, R Zhao… - Advances in neural …, 2020 - proceedings.neurips.cc
In this paper, we explore the limits of Microsoft Floating Point (MSFP), a new class of
datatypes developed for production cloud-scale inferencing on custom hardware. Through …
datatypes developed for production cloud-scale inferencing on custom hardware. Through …
Stable and low-precision training for large-scale vision-language models
M Wortsman, T Dettmers… - Advances in …, 2023 - proceedings.neurips.cc
We introduce new methods for 1) accelerating and 2) stabilizing training for large language-
vision models. 1) For acceleration, we introduce SwitchBack, a linear layer for int8 quantized …
vision models. 1) For acceleration, we introduce SwitchBack, a linear layer for int8 quantized …
Dynamic aggregation for heterogeneous quantization in federated learning
Communication is widely known as the primary bottleneck of federated learning, and
quantization of local model updates before uploading to the parameter server is an effective …
quantization of local model updates before uploading to the parameter server is an effective …
Tensordash: Exploiting sparsity to accelerate deep neural network training
TensorDash is a hardware-based technique that enables data-parallel MAC units to take
advantage of sparsity in their input operand streams. When used to compose a hardware …
advantage of sparsity in their input operand streams. When used to compose a hardware …