Search for: All records

Machine learning has shown incredibly potential in many fields of application ranging from ChatGPT and Bard to Tesla’s autonomous vehicles. These ML models require vast amounts of data and communications overhead in order to be effective. In this paper we propose a communication-efficient time series forecasting model combining the most recent advancements in MetaFormer architecture implemented across a federated series of learning nodes. The time series prediction performance and communication overhead cost of the distributed model is compared against a similar centralized model and shown to have parity in performance while consuming much lower data rates during training. 

Polyanskiy [1] proposed a framework for the MAC problem with a large number of users, where users employ a common codebook in the finite blocklength regime. In this work, we extend [1] to the case when the number of active users is random and there is also a delay constraint. We first define a random-access channel and derive the general converse bound. Our bound captures the basic tradeoff between the required energy and the delay constraint. Then we propose an achievable bound for block transmission. In this case, all packets are transmitted in the second half of the block to avoid interference. We then study treating interference as noise (TIN) with both single user and multiple users. Last, we derive an achievable bound for the packet splitting model, which allows users to split each packet into two parts with different blocklengths. Our numerical results indicate that, when the delay is large, TIN is effective; on the other hand, packet splitting outperforms as the delay decreases. 

Two-branch network architecture has shown its efficiency and effectiveness in real-time semantic segmentation tasks. However, direct fusion of high-resolution details and low-frequency context has the drawback of detailed features being easily overwhelmed by surrounding contextual information. This overshoot phenomenon limits the improvement of the segmentation accuracy of existing two-branch mod- els. In this paper, we make a connection between Convolutional Neural Networks (CNN) and Proportional-Integral-Derivative (PID) controllers and reveal that a two-branch network is equivalent to a Proportional-Integral (PI) controller, which inherently suffers from similar overshoot issues. To alleviate this problem, we propose a novel three- branch network architecture: PIDNet, which contains three branches to parse detailed, context and boundary information, respectively, and employs boundary attention to guide the fusion of detailed and context branches. Our family of PIDNets achieve the best trade-off between inference speed and accuracy and their accuracy surpasses all the existing models with similar inference speed on the Cityscapes and CamVid datasets. Specifically, PIDNet-S achieves 78.6% mIOU with inference speed of 93.2 FPS on Cityscapes and 80.1% mIOU with speed of 153.7 FPS on CamVid. 

This paper investigates basic trade-offs between energy and delay in wireless communication systems using finite blocklength theory. We first assume that data arrive in constant stream of bits, which are put into packets and transmitted over a communications link. Our results show that depending on exactly how energy is measured, in general energy depends on sqrt{d^{-1}} or sqrt{d^{-1}log d}, where d is the delay. This means that the energy decreases quite slowly with increasing delay. Furthermore, to approach the absolute minimum of -1.59 dB on energy, bandwidth has to increase very rapidly, much more than what is predicted by infinite blocklength theory. We then consider the scenario when data arrive stochastically in packets and can be queued. We devise a scheduling algorithm based on finite blocklength theory and develop bounds for the energy-delay performance. Our results again show that the energy decreases quite slowly with increasing delay.