The occurrence of manufacturing defects in wind turbine blade (WTB) production can result in significant increases in operation and maintenance costs and lead to severe and disastrous consequences. Therefore, inspection during the manufacturing process is crucial to ensure consistent fabrication of composite materials. Non-contact sensing techniques, such as Frequency Modulated Continuous Wave (FMCW) radar, are becoming increasingly popular as they offer a full view of these complex structures during curing. In this paper, we enhance the quality assurance of manufacturing utilizing FMCW radar as a non-destructive sensing modality. Additionally, a novel anomaly detection pipeline is developed that offers the following advantages: (1) We use the analytic representation of the Intermediate Frequency signal of the FMCW radar as a feature to disentangle material-specific and round-trip delay information from the received wave. (2) We propose a novel anomaly detection methodology called focus Support Vector Data Description (focus-SVDD). This methodology involves defining the limit boundaries of the dataset after removing healthy data features, thereby focusing on the attributes of anomalies. (3) The proposed method employs a complex-valued autoencoder to remove healthy features and we introduces a new activation function called Exponential Amplitude Decay (EAD). EAD takes advantage of the Rayleigh distribution, which characterizes an instantaneous amplitude signal. The effectiveness of the proposed method is demonstrated through its application to collected data, where it shows superior performance compared to other state-of-the-art unsupervised anomaly detection methods. This method is expected to make a significant contribution not only to structural health monitoring but also to the field of deep complex-valued data processing and SVDD application.

This paper presents FDNet: a Focal Decomposed Network for efficient, robust and practical time series forecasting. We break away from conventional deep time series forecasting formulas which obtain prediction results from universal feature maps of input sequences. In contrary, FDNet neglects universal correlations of input elements and only extracts fine-grained local features from input sequence. We show that: (1) Deep time series forecasting with only fine-grained local feature maps of input sequence is feasible upon theoretical basis. (2) By abandoning global coarse-grained feature maps, FDNet overcomes distribution shift problem caused by changing dynamics of time series which is common in real-world applications. (3) FDNet is not dependent on any inductive bias of time series except basic auto-regression, making it general and practical. Moreover, we propose focal input sequence decomposition method which decomposes input sequence in a focal manner for efficient and robust forecasting when facing Long Sequence Time series Input (LSTI) problem. FDNet achieves competitive forecasting performances on six real-world benchmarks and reduces prediction MSE by 38.4% on average compared with other thirteen SOTA baselines. The source code is available at https://github.com/OrigamiSL/FDNet.

In this paper, we propose the Whiplash Inertial Gradient dynamics, a closed-loop optimization method that utilises gradient information, to find the minima of a cost function in finite-dimensional settings. We introduce the symplectic asymptotic convergence analysis for the Whiplash system for convex functions. We also introduce relaxation sequences to explain the non-classical nature of the algorithm and an exploring heuristic variant of the Whiplash algorithm to escape saddle points, deterministically. We study the algorithm's performance for various costs and provide a practical methodology for analyzing convergence rates using integral constraint bounds and a novel Lyapunov rate method. Our results demonstrate polynomial and exponential rates of convergence for quadratic cost functions. In this paper, we study the continuous optimization of finite-dimensional, unconstrained problems. We revisit classical optimization theories at the heart of popular deep learning algorithms. These problems arise in fields such as deep learning, economics, and physics. With stochastic gradient approaches facing information bottlenecks [1], it is important to revisit deterministic approaches to understand ways to improve modern optimization tools from both practical and theoretical perspectives. Secondorder methods which are significantly faster [2] often tend to be computationally infeasible.

We proposes a simple deep learning architecture combining elements of Inception, ResNet and Xception networks. Four new datasets were used for classification with both small and large training samples. Results in terms of classification accuracy suggests improved performance by proposed architecture in comparison to Bayesian optimised 2D-CNN with small training samples. Comparison of results using small training sample with Indiana Pines hyperspectral dataset suggests comparable or better performance by proposed architecture than nine reported works using different deep learning architectures. In spite of achieving high classification accuracy with limited training samples, comparison of classified image suggests different land cover classes are assigned to same area when compared with the classified image provided by the model trained using large training samples with all datasets.

Generative Flow Networks or GFlowNets are related to Monte-Carlo Markov chain methods (as they sample from a distribution specified by an energy function), reinforcement learning (as they learn a policy to sample composed objects through a sequence of steps), generative models (as they learn to represent and sample from a distribution) and amortized variational methods (as they can be used to learn to approximate and sample from an otherwise intractable posterior, given a prior and a likelihood). They are trained to generate an object $x$ through a sequence of steps with probability proportional to some reward function $R(x)$ (or $\exp(-\mathcal{E}(x))$ with $\mathcal{E}(x)$ denoting the energy function), given at the end of the generative trajectory. Like for other RL settings where the reward is only given at the end, the efficiency of training and credit assignment may suffer when those trajectories are longer. With previous GFlowNet work, no learning was possible from incomplete trajectories (lacking a terminal state and the computation of the associated reward). In this paper, we consider the case where the energy function can be applied not just to terminal states but also to intermediate states. This is for example achieved when the energy function is additive, with terms available along the trajectory. We show how to reparameterize the GFlowNet state flow function to take advantage of the partial reward already accrued at each state. This enables a training objective that can be applied to update parameters even with incomplete trajectories. Even when complete trajectories are available, being able to obtain more localized credit and gradients is found to speed up training convergence, as demonstrated across many simulations.

Quantum computers are known to provide speedups over classical state-of-the-art machine learning methods in some specialized settings. For example, quantum kernel methods have been shown to provide an exponential speedup on a learning version of the discrete logarithm problem. Understanding the generalization of quantum models is essential to realizing similar speedups on problems of practical interest. Recent results demonstrate that generalization is hindered by the exponential size of the quantum feature space. Although these results suggest that quantum models cannot generalize when the number of qubits is large, in this paper we show that these results rely on overly restrictive assumptions. We consider a wider class of models by varying a hyperparameter that we call quantum kernel bandwidth. We analyze the large-qubit limit and provide explicit formulas for the generalization of a quantum model that can be solved in closed form. Specifically, we show that changing the value of the bandwidth can take a model from provably not being able to generalize to any target function to good generalization for well-aligned targets. Our analysis shows how the bandwidth controls the spectrum of the kernel integral operator and thereby the inductive bias of the model. We demonstrate empirically that our theory correctly predicts how varying the bandwidth affects generalization of quantum models on challenging datasets, including those far outside our theoretical assumptions. We discuss the implications of our results for quantum advantage in machine learning.

Rearranging and manipulating deformable objects such as cables, fabrics, and bags is a long-standing challenge in robotic manipulation. The complex dynamics and high-dimensional configuration spaces of deformables, compared to rigid objects, make manipulation difficult not only for multi-step planning, but even for goal specification. Goals cannot be as easily specified as rigid object poses, and may involve complex relative spatial relations such as "place the item inside the bag". In this work, we develop a suite of simulated benchmarks with 1D, 2D, and 3D deformable structures, including tasks that involve image-based goal-conditioning and multi-step deformable manipulation. We propose embedding goal-conditioning into Transporter Networks, a recently proposed model architecture for learning robotic manipulation that rearranges deep features to infer displacements that can represent pick and place actions. In simulation and in physical experiments, we demonstrate that goal-conditioned Transporter Networks enable agents to manipulate deformable structures into flexibly specified configurations without test-time visual anchors for target locations. We also significantly extend prior results using Transporter Networks for manipulating deformable objects by testing on tasks with 2D and 3D deformables. Supplementary material is available at https://berkeleyautomation.github.io/bags/.

In recent years, the gap between Deep Learning (DL) methods and analytical or numerical approaches in scientific computing is tried to be filled by the evolution of Physics-Informed Neural Networks (PINNs). However, still, there are many complications in the training of PINNs and optimal interleaving of physical models. Here, we introduced the concept of Physical Activation Functions (PAFs). This concept offers that instead of using general activation functions (AFs) such as ReLU, tanh, and sigmoid for all the neurons, one can use generic AFs that their mathematical expression is inherited from the physical laws of the investigating phenomena. The formula of PAFs may be inspired by the terms in the analytical solution of the problem. We showed that the PAFs can be inspired by any mathematical formula related to the investigating phenomena such as the initial or boundary conditions of the PDE system. We validated the advantages of PAFs for several PDEs including the harmonic oscillations, Burgers, Advection-Convection equation, and the heterogeneous diffusion equations. The main advantage of PAFs was in the more efficient constraining and interleaving of PINNs with the investigating physical phenomena and their underlying mathematical models. This added constraint significantly improved the predictions of PINNs for the testing data that was out-of-training distribution. Furthermore, the application of PAFs reduced the size of the PINNs up to 75% in different cases. Also, the value of loss terms was reduced by 1 to 2 orders of magnitude in some cases which is noteworthy for upgrading the training of the PINNs. The iterations required for finding the optimum values were also significantly reduced. It is concluded that using the PAFs helps in generating PINNs with less complexity and much more validity for longer ranges of prediction.

Reinforcement learning (RL) problems can be challenging without well-shaped rewards. Prior work on provably efficient RL methods generally proposes to address this issue with dedicated exploration strategies. However, another way to tackle this challenge is to reformulate it as a multi-task RL problem, where the task space contains not only the challenging task of interest but also easier tasks that implicitly function as a curriculum. Such a reformulation opens up the possibility of running existing multi-task RL methods as a more efficient alternative to solving a single challenging task from scratch. In this work, we provide a theoretical framework that reformulates a single-task RL problem as a multi-task RL problem defined by a curriculum. Under mild regularity conditions on the curriculum, we show that sequentially solving each task in the multi-task RL problem is more computationally efficient than solving the original single-task problem, without any explicit exploration bonuses or other exploration strategies. We also show that our theoretical insights can be translated into an effective practical learning algorithm that can accelerate curriculum learning on simulated robotic tasks.

Salimans et al. (2017) have shown that such optimization methods may We present a representation-driven framework for cause high variance updates in long horizon problems, while reinforcement learning. By representing policies Tessler et al. (2019) have shown possible convergence to as estimates of their expected values, we leverage suboptimal solutions in continuous regimes. Moreover, policy techniques from contextual bandits to guide exploration search methods are commonly sample inefficient, particularly and exploitation. Particularly, embedding in hard exploration problems, as policy gradient a policy network into a linear feature space allows methods usually converge to areas of high reward, without us to reframe the exploration-exploitation sacrificing exploration resources to achieve a far-reaching problem as a representation-exploitation problem, sparse reward.