As a substantial amount of multivariate time series data is being produced by the complex systems in Smart Manufacturing, improved anomaly detection frameworks are needed to reduce the operational risks and the monitoring burden placed on the system operators. However, building such frameworks is challenging, as a sufficiently large amount of defective training data is often not available and frameworks are required to capture both the temporal and contextual dependencies across different time steps while being robust to noise. In this paper, we propose an unsupervised Attention-based Convolutional Long Short-Term Memory (ConvLSTM) Autoencoder with Dynamic Thresholding (ACLAE-DT) framework for anomaly detection and diagnosis in multivariate time series. The framework starts by pre-processing and enriching the data, before constructing feature images to characterize the system statuses across different time steps by capturing the inter-correlations between pairs of time series. Afterwards, the constructed feature images are fed into an attention-based ConvLSTM autoencoder, which aims to encode the constructed feature images and capture the temporal behavior, followed by decoding the compressed knowledge representation to reconstruct the feature images input. The reconstruction errors are then computed and subjected to a statistical-based, dynamic thresholding mechanism to detect and diagnose the anomalies. Evaluation results conducted on real-life manufacturing data demonstrate the performance strengths of the proposed approach over state-of-the-art methods under different experimental settings.
Existing Simultaneous Localization and Mapping (SLAM) approaches are limited in their scalability due to growing map size in long-term robot operation. Moreover, processing such maps for localization and planning tasks leads to the increased computational resources required onboard. To address the problem of memory consumption in long-term operation, we develop a novel real-time SLAM algorithm, MeSLAM, that is based on neural field implicit map representation. It combines the proposed global mapping strategy, including neural networks distribution and region tracking, with an external odometry system. As a result, the algorithm is able to efficiently train multiple networks representing different map regions and track poses accurately in large-scale environments. Experimental results show that the accuracy of the proposed approach is comparable to the state-of-the-art methods (on average, 6.6 cm on TUM RGB-D sequences) and outperforms the baseline, iMAP$^*$. Moreover, the proposed SLAM approach provides the most compact-sized maps without details distortion (1.9 MB to store 57 m$^3$) among the state-of-the-art SLAM approaches.
Safe navigation in real-time is challenging because engineers need to work with uncertain vehicle dynamics, variable external disturbances, and imperfect controllers. A common safety strategy is to inflate obstacles by hand-defined margins. However, arbitrary static margins often fail in more dynamic scenarios, and using worst-case assumptions is overly conservative for most settings where disturbances over time. In this work, we propose a middle ground: safety margins that adapt on-the-fly. In an offline phase, we use Monte Carlo simulations to pre-compute a library of safety margins for multiple levels of disturbance uncertainties. Then, at runtime, our system estimates the current disturbance level to query the associated safety margins that best trades off safety and performance. We validate our approach with extensive simulated and real-world flight tests. We show that our adaptive method significantly outperforms static margins, allowing the vehicle to operate up to 1.5 times faster than worst-case static margins while maintaining safety. Video: https://youtu.be/SHzKHSUjdUU
Recent techniques to solve photorealistic style transfer within deep convolutional neural networks (CNNs) generally require intensive training from large-scale datasets, thus having limited applicability and poor generalization ability to unseen images or styles. To overcome this, we propose a novel framework, dubbed Deep Translation Prior (DTP), to accomplish photorealistic style transfer through test-time training on given input image pair with untrained networks, which learns an image pair-specific translation prior and thus yields better performance and generalization. Tailored for such test-time training for style transfer, we present novel network architectures, with two sub-modules of correspondence and generation modules, and loss functions consisting of contrastive content, style, and cycle consistency losses. Our framework does not require offline training phase for style transfer, which has been one of the main challenges in existing methods, but the networks are to be solely learned during test-time. Experimental results prove that our framework has a better generalization ability to unseen image pairs and even outperforms the state-of-the-art methods.
Adversarial training is a widely used strategy for making neural networks resistant to adversarial perturbations. For a neural network of width $m$, $n$ input training data in $d$ dimension, it takes $\Omega(mnd)$ time cost per training iteration for the forward and backward computation. In this paper we analyze the convergence guarantee of adversarial training procedure on a two-layer neural network with shifted ReLU activation, and shows that only $o(m)$ neurons will be activated for each input data per iteration. Furthermore, we develop an algorithm for adversarial training with time cost $o(m n d)$ per iteration by applying half-space reporting data structure.
Since the recognition in the early nineties of the vanishing/exploding (V/E) gradient issue plaguing the training of neural networks (NNs), significant efforts have been exerted to overcome this obstacle. However, a clear solution to the V/E issue remained elusive so far. In this manuscript a new architecture of NN is proposed, designed to mathematically prevent the V/E issue to occur. The pursuit of approximate dynamical isometry, i.e. parameter configurations where the singular values of the input-output Jacobian are tightly distributed around 1, leads to the derivation of a NN's architecture that shares common traits with the popular Residual Network model. Instead of skipping connections between layers, the idea is to filter the previous activations orthogonally and add them to the nonlinear activations of the next layer, realising a convex combination between them. Remarkably, the impossibility for the gradient updates to either vanish or explode is demonstrated with analytical bounds that hold even in the infinite depth case. The effectiveness of this method is empirically proved by means of training via backpropagation an extremely deep multilayer perceptron of 50k layers, and an Elman NN to learn long-term dependencies in the input of 10k time steps in the past. Compared with other architectures specifically devised to deal with the V/E problem, e.g. LSTMs for recurrent NNs, the proposed model is way simpler yet more effective. Surprisingly, a single layer vanilla RNN can be enhanced to reach state of the art performance, while converging super fast; for instance on the psMNIST task, it is possible to get test accuracy of over 94% in the first epoch, and over 98% after just 10 epochs.
Lidar-based simultaneous localization and mapping (SLAM) approaches have obtained considerable success in autonomous robotic systems. This is in part owing to the high-accuracy of robust SLAM algorithms and the emergence of new and lower-cost lidar products. This study benchmarks current state-of-the-art lidar SLAM algorithms with a multi-modal lidar sensor setup showcasing diverse scanning modalities (spinning and solid-state) and sensing technologies, and lidar cameras, mounted on a mobile sensing and computing platform. We extend our previous multi-modal multi-lidar dataset with additional sequences and new sources of ground truth data. Specifically, we propose a new multi-modal multi-lidar SLAM-assisted and ICP-based sensor fusion method for generating ground truth maps. With these maps, we then match real-time pointcloud data using a natural distribution transform (NDT) method to obtain the ground truth with full 6 DOF pose estimation. This novel ground truth data leverages high-resolution spinning and solid-state lidars. We also include new open road sequences with GNSS-RTK data and additional indoor sequences with motion capture (MOCAP) ground truth, complementing the previous forest sequences with MOCAP data. We perform an analysis of the positioning accuracy achieved with ten different SLAM algorithm and lidar combinations. We also report the resource utilization in four different computational platforms and a total of five settings (Intel and Jetson ARM CPUs). Our experimental results show that current state-of-the-art lidar SLAM algorithms perform very differently for different types of sensors. More results, code, and the dataset can be found at: \href{https://github.com/TIERS/tiers-lidars-dataset-enhanced}{github.com/TIERS/tiers-lidars-dataset-enhanced.
In this letter, we study the simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) assisted unmanned aerial vehicle (UAV) communications. Our goal is to maximize the sum rate of all users by jointly optimizing the STAR-RIS's beamforming vectors, the UAV's trajectory and power allocation. We decompose the formulated non-convex problem into three subproblems and solve them alternately to obtain the solution. Simulations show that: 1) the STAR-RIS achieves a higher sum rate than traditional RIS; 2) to exploit the benefits of STAR-RIS, the UAV's trajectory is closer to STAR-RIS than that of RIS; 3) the energy splitting for reflection and transmission highly depends on the real-time trajectory of UAV.
Traditional solvers for tackling combinatorial optimization (CO) problems are usually designed by human experts. Recently, there has been a surge of interest in utilizing Deep Learning, especially Deep Reinforcement Learning, to automatically learn effective solvers for CO. The resultant new paradigm is termed Neural Combinatorial Optimization (NCO). However, the advantages and disadvantages of NCO over other approaches have not been well studied empirically or theoretically. In this work, we present a comprehensive comparative study of NCO solvers and alternative solvers. Specifically, taking the Traveling Salesman Problem as the testbed problem, we assess the performance of the solvers in terms of five aspects, i.e., effectiveness, efficiency, stability, scalability and generalization ability. Our results show that in general the solvers learned by NCO approaches still fall short of traditional solvers in nearly all these aspects. A potential benefit of the former would be their superior time and energy efficiency on small-size problem instances when sufficient training instances are available. We hope this work would help better understand the strengths and weakness of NCO, and provide a comprehensive evaluation protocol for further benchmarking NCO approaches against other approaches.
We propose Beat Transformer, a novel Transformer encoder architecture for joint beat and downbeat tracking. Different from previous models that track beats solely based on the spectrogram of an audio mixture, our model deals with demixed spectrograms with multiple instrument channels. This is inspired by the fact that humans perceive metrical structures from richer musical contexts, such as chord progression and instrumentation. To this end, we develop a Transformer model with both time-wise attention and instrument-wise attention to capture deep-buried metrical cues. Moreover, our model adopts a novel dilated self-attention mechanism, which achieves powerful hierarchical modelling with only linear complexity. Experiments demonstrate a significant improvement in demixed beat tracking over the non-demixed version. Also, Beat Transformer achieves up to 4% point improvement in downbeat tracking accuracy over the TCN architectures. We further discover an interpretable attention pattern that mirrors our understanding of hierarchical metrical structures.