In recent years, multi-classifier learning is of significant interest in industrial and economic fields. Moreover, neural network is a popular approach in multi-classifier learning. However, the accuracies of neural networks are often limited by their loss functions. For this reason, we design a novel cross entropy loss function, named MPCE, which based on the maximum probability in predictive results. In this paper, we first analyze the difference of gradients between MPCE and the cross entropy loss function. Then, we propose the gradient update algorithm based on MPCE. In the experimental part of this paper, we utilize four groups of experiments to verify the performance of the proposed algorithm on six public datasets. The first group of experimental results show that the proposed algorithm converge faster than the algorithms based on other loss functions. Moreover, the results of the second group show that the proposed algorithm obtains the highest training and test accuracy on the six datasets, and the proposed algorithm perform better than others when class number changing on the sensor dataset. Furthermore, we use the model of convolutional neural network to implement the compared methods on the mnist dataset in the fourth group of experiments. The results show that the proposed algorithm has the highest accuracy among all executed methods.

Many problems in the Internet of Things (IoT) can be regarded as online optimization problems. For this reason, an online constrained problem in IoT is considered in this paper, where the cost functions change over time. To solve this problem, many projected online optimization algorithms have been wildly used. However, the projections of these algorithms become prohibitive in problems involving high-dimensional parameters and massive data. To address this issue, we propose a Frank-Wolfe Adam online learning algorithm called FWAdam, which uses a Frank-Wolfe method to eschew costly projection operations. Furthermore, we first give the convergence analysis of the FWAdam algorithm, and prove its regret bound to O(T3/4) when cost functions are convex, where T is a time horizon. Finally, we present simulated experiments on two datasets to validate our theoretical results.

Visual-Inertial Odometry (VIO) provides a robust localization solution for Simultaneous Localization and Mapping (SLAM) systems. Self-supervised VIO, a leading approach, has the advantage of not requiring extensive ground-truth labels. Regrettably, this method still poses challenges for robotic applications, particularly unmanned aerial vehicles (UAVs), due to its computational complexity arising from inadequate model designs. To address this bottleneck, we introduce BotVIO (where “Bot” refers to “robotics”), a Transformer-based self-supervised VIO model, offering an excellent solution to alleviate computational burdens for robotics. Our lightweight backbone combines shallow CNNs with spatial-temporal-enhanced Transformers to replace conventional architectures, while the minimalist cross-fusion module uses single-layer cross-attention to enhance multimodal interaction. Extensive experiments show that, during pose estimation, BotVIO achieves a remarkable 70.37% reduction in trainable parameters and a 74.85% decrease in inference speed, reaching up to 57.80 fps on an NVIDIA Jetson NX (10W&2CORE), while improving pose accuracy and robustness. For the benefit of the community, we make public the source code.¹¹https://github.com/wenhuiwei-ustc/BotVIO

This paper investigates the data transport problem for reporting delay-sensitive events in wireless sensor-actuator networks (WSANs). We specifically tailor the protocol design according to the features of WSANs and propose POWER-SPEED, a real-time data transport protocol for WSANs to achieve energy-efficient data transport for delay-sensitive event reporting. In POWER-SPEED, sensor nodes select the next-hop neighbor to actuators according to the spatio-temporal historic data of the upstream QoS condition, which completely avoids control packets. With an adaptive transmitter power control scheme, POWER-SPEED conveys packets in an energy-efficient manner while maintaining soft real-time packet transport. It thus reduces the energy consumption of data transport while ensuring the QoS requirement in timeliness domain. We demonstrate the effectiveness of POWER-SPEED through simulations with NS2.

AdaBelief, one of the current best optimizers, demonstrates superior generalization ability over the popular Adam algorithm by viewing the exponential moving average of observed gradients. AdaBelief is theoretically appealing in which it has a data-dependent O(√T) regret bound when objective functions are convex, where T is a time horizon. It remains, however, an open problem whether the convergence rate can be further improved without sacrificing its generalization ability. To this end, we make the first attempt in this work and design a novel optimization algorithm called FastAdaBelief that aims to exploit its strong convexity in order to achieve an even faster convergence rate. In particular, by adjusting the step size that better considers strong convexity and prevents fluctuation, our proposed FastAdaBelief demonstrates excellent generalization ability and superior convergence. As an important theoretical contribution, we prove that FastAdaBelief attains a data-dependent O(log T) regret bound, which is substantially lower than AdaBelief in strongly convex cases. On the empirical side, we validate our theoretical analysis with extensive experiments in scenarios of strong convexity and nonconvexity using three popular baseline models. Experimental results are very encouraging: FastAdaBelief converges the quickest in comparison to all mainstream algorithms while maintaining an excellent generalization ability, in cases of both strong convexity or nonconvexity. FastAdaBelief is, thus, posited as a new benchmark model for the research community.