Jiahuan Zhang, Shunwen Bai, Tianheng Wang, KaiWen Guo, Zijia Song, Hanqing WU, Guozheng Rao, Kai Han, Kaicheng Yu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Multilayer Perceptron (MLP) is a simple practice of Neural Network (NN) and the cornerstone of research and development of deep learning. Each neuron is connected to all neurons in the previous layer and implements a non-linear mapping through activation functions. MLP can learn complex non-linear relationships among features through the superposition of multiple hidden layers, but it still cannot discover the inherent strong correlation among features. The reason is that each neuron uses a simple weighted summation method to organize all the neurons in the previous layer. Inspired by quantum theory, this paper builds a non-linear NN layer that can mine strong correlations among features based on multi-body quantum systems, and then constructs a multi-layer perceptron, called Quantum-inspired MLP (QiMLP). It is conceivable that QiMLP will have important inspirational significance in reshaping machine learning, deep learning and large language models. We theoretically analyzed the basis for QiMLP to mine strong correlations among features, and implemented experiments on multiple classic deep learning datasets. Experimental results verify that QiMLP not only learns strong correlations among features, but also significantly reduces the number of parameters with hundreds of times improvement.

Jiahuan Zhang, Tianheng Wang, Ziyi Huang, Yulong Wu, Hanqing Wu, DongbaiChen DongbaiChen, Linfeng Song, Yue Zhang, Guozheng Rao, Kaicheng Yu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

In this paper, we address the problem of estimating the rotational extrinsics, as well as the scale factors of two gyroscopes rigidly mounted on the same device. In particular, we formulate the problem as a least-squares minimization and introduce a direct algorithm that computes the estimated quantities without any iterations, hence avoiding local minima and improving efficiency. Furthermore, we show that the rotational extrinsics are observable while the scale factors can be determined up to a global scale for general configurations of the gyroscopes. To this end, we also study special placements of the gyroscopes where a pair, or all, of their axes are parallel and analyze their impact on the scale factors’ observability. Lastly, we evaluate our algorithm in simulation and real-world experiments to assess its performance as a function of key motion and sensor characteristics.

In this paper, we focus on buffer-aided wireless powered Internet of Things (IoTs) comprising of one wireless access point (AP) and multiple devices, where the AP provides energy to all devices via downlink radio frequency (RF) energy beams. All devices utilize the harvested energy to transmit their data to the AP in a time-division multiple access (TDMA) manner. Every device is assumed to be provisioned with energy storage and data buffer to store the collected energy from the AP and its data, respectively. The problem of minimizing the long-term average age of information (AoI) of the system is formulated in this paper. By solving the problem under the Lyapunov optimization framework, the AoI-aware adaptive transmission scheme is obtained, in which downlink RF energy beamforming, downlink energy transfer and uplink access, as well as transmit power and transmission rate by every device, will be jointly adjusted in order to minimize average weightede AoI according to the underlying channel state information (CSI), the buffer state information (BSI), the energy-consumption status information (ESI) of all terminals, as well as the AoI status information (ASI). Our analysis unveils that, the status update rate at devices has a significant impact on the achievable AoI performance, and the minimum average weighted AoI can only be realized at a reasonable status update rate, which is neither too high nor too low. Moreover, flexible AoI-aware scheme can be realized by adjusting either the AoI priority level or the AoI weighting coefficient.

Network navigation is a promising paradigm for enabling location awareness in dynamic wireless networks. A wireless navigation network consists of agents (mobile with unknown locations) and anchors (possibly mobile with known locations). Agents can estimate their locations based on inter-and intra- node measurements as well as prior knowledge. With limited wireless resources, the key to achieve high navigation accuracy is to maximize the benefits of agents' channel usage. Therefore, it is critical to design scheduling algorithms that adaptively determine with whom and when an agent should perform inter-node measurements to achieve both high navigation accuracy and efficient channel usage. This paper develops a framework for the design of distributed scheduling algorithms in asynchronous wireless navigation networks, under which the algorithm parameters are optimized based on the evolution of agents' localization errors. Results show that the proposed algorithms lead to high-accuracy, efficient, and flexible network navigation.

Network navigation is a promising paradigm for providing location awareness in wireless environments, where nodes estimate their locations based on sensor measurements and prior knowledge. In the presence of limited wireless resources, only a subset rather than all of the node pairs can perform inter-node measurements. The procedure of selecting node pairs at different times for inter-node measurements, referred to as network scheduling, affects the evolution of the localization errors. Thus, it is crucial to design efficient scheduling strategies for network navigation. This paper introduces situation-aware scheduling that exploits network states to select measurement pairs, and develops a framework to characterize the effects of scheduling strategies and of network settings on the error evolution. In particular, both sufficient and necessary conditions for the boundedness of the error evolution are provided. Furthermore, opportunistic and random situation-aware scheduling strategies are proposed, and bounds on the corresponding time-averaged network localization errors are derived. These strategies are shown to be optimal in terms of the error scaling with the number of agents. Finally, the reduction of the error scaling by increasing the number of simultaneous measurement pairs is quantified.

Network navigation is a promising paradigm for providing accurate location-awareness in wireless environments, where mobile nodes estimate their locations based on inter- and intra-node measurements. In the presence of limited wireless resources, only a subset rather than all of the node pairs can perform inter-node measurements. Therefore, it is crucial to design efficient scheduling algorithms for selecting node pairs at different times for inter-node measurements. This paper develops a framework for the design of scheduling algorithms based on random access for network navigation. The proposed algorithms are suitable for practical operation of wireless navigation networks due to their distributed nature, and the optimized access probabilities of the agents lead to significant performance improvement.

Network navigation is an emerging paradigm for enabling location-awareness in wireless networks. The limited wireless resource calls for scheduling algorithms to select node pairs for inter-node measurements. The design of efficient scheduling algorithms relies on an understanding of the underlying localization error evolution. This paper unifies the error evolution analysis in both the non-Bayesian and Bayesian cases. We first develop a unified recursive equation for the error evolution in the two cases. We then provide sufficient conditions for the boundedness of the error evolution. These conditions are more general than our previous results and can be used to derive error bounds for different scheduling algorithms. Furthermore, we show the the reduction of the error scaling with respect to the number of agents by increasing the multiplexing factor. These results provide insights for the efficient operation of wireless navigation networks.

Location-awareness plays a key role in various applications in future wireless networks. In GPS-challenged environments, location-awareness can be achieved via wireless navigation networks, which call for an efficient scheduling algorithm to optimize the navigation performance under communication constraints. In this paper, we develop a general framework for the design and analysis of scheduling algorithms for navigation networks with multiple measurement pairs per time slot. In particular, we provide sufficient and necessary conditions for the stability of the error evolution, and derive bounds on the time-averaged network localization errors (NLEs) for opportunistic and random scheduling. Furthermore, we show that the two scheduling algorithms are optimal in terms of the error scaling with respect to the number of agents, and we quantify the performance gain from measurement pair selections exploiting the network states. These results provide guidelines for designing efficient scheduling algorithms for network navigation.

Wireless navigation networks enable location-awareness in GPS-challenged environments. For such networks, scheduling algorithms are needed to improve the navigation accuracy through measurement pair selections under limited communication resource. In this paper, we develop an analytical framework to determine the location error evolution for different scheduling algorithms and network settings. Under this framework, we provide sufficient conditions for the stability of the location error evolution, and we quantify the time-averaged network location errors (NLEs) for scheduling algorithms with and without exploiting the network states. Furthermore, we show the optimality of the proposed scheduling algorithms in terms of the error scaling with respect to the agent density. These results provide fundamental insights into the effects of scheduling algorithms and network settings on the location error evolution, leading to efficient scheduling algorithms for navigation networks.

In cooperative localization networks, nodes estimate their locations through inter-node ranging, location information exchange, and information fusion. These operations cause packet collision, large communication overhead, and high computational complexity, which can be alleviated by proper scheduling techniques. In this paper, we design distributed scheduling algorithms for cooperative localization networks, which improve the efficiency of localization through neighbor selection and collision control. Then, we determine the evolution of each node's location error through cooperation and movement using Fisher information analysis. Furthermore, we analyze the convergence of the location error under the proposed scheduling algorithm. Simulation results show that the efficiency of localization is significantly improved by using the proposed scheduling algorithm.

The combination of relay and Multiple-Input Multiple-Output (MIMO) techniques can significantly benefit the signal transmission in cellular networks. In this work, we propose a novel relaying scheme for the relay assisted MIMO broadcast channel including user classification and preceding. The users are firstly classified as Direct Link (DL) users and Relay Link (RL) users according to their channel conditions. After classification, both DL and RL users may coexist in the system, and Zero-Forcing (ZF) preceding is employed at both the BS and the relay for interference cancellation. Analysis and simulation results show that the proposed relaying scheme can effectively combat the large scale fading of the channels and provide significant performance gain compared with the conventional MIMO relaying schemes without user classification.