The district heating network (DHN) can provide flexibility for electric power system (EPS) to accommodate power because of its slow dynamic and thermal energy storage characteristics. However, the traditional flexibility quantifications neglect the thermal temporal-spatial dynamic propagation and uncertainty parameters of DHN, resulting in inaccurate assessment of available flexibility capabilities. To address this issue, this article proposes a dynamic flexibility quantification method. First, three flexibility metrics including capacity, amplitude, and duration of power integration are modeled by simultaneously considering the temperature temporal and spatial dynamic propagation. Then, the discretized criteria are designed to reasonably linearize the temporal and spatial dynamic variables in the metrics. Thus, the evolution of flexibility metrics over time and space can be captured. Furthermore, the uncertainty parameters (e.g., mass flow, thermal resistance) in the metrics are modeled to account for their impact on the flexibility results. To this end, the maximum entropy principle combined with the probabilistic cumulant is developed to construct the probability distributions of metrics. With these effects, the flexibility provided by DHN can be elaborately quantified. Finally, case studies and simulation analysis are carried out on China Luhua network and Denmark 61-node network to verify the effectiveness of the proposed quantification method.

We study the problem of symbolic music generation (e.g., generating piano rolls), with a technical focus on non-differentiable rule guidance. Musical rules are often expressed in symbolic form on note characteristics, such as note density or chord progression, many of which are non-differentiable which pose a challenge when using them for guided diffusion. We propose Stochastic Control Guidance (SCG), a novel guidance method that only requires forward evaluation of rule functions that can work with pre-trained diffusion models in a plug-and-play way, thus achieving training-free guidance for non-differentiable rules for the first time. Additionally, we introduce a latent diffusion architecture for symbolic music generation with high time resolution, which can be composed with SCG in a plug-and-play fashion. Compared to standard strong baselines in symbolic music generation, this framework demonstrates marked advancements in music quality and rule-based controllability, outperforming current state-of-the-art generators in a variety of settings. For detailed demonstrations, code and model checkpoints, please visit our [project website](https://scg-rule-guided-music.github.io/).

With renewable energy becoming more and more important, the heat and electricity integrated energy system (HE-IES) has been widely used because of its potential benefit in accommodating renewable energy. This work for the first time formulated a dynamic optimal energy flow (OEF) problem for the HE-IES with hybrid AC/DC sources. Firstly, to effectively and accurately assess the operation state of district heating network (DHN), the temperature dynamics are focused and solved with a set of difference equations where an adaptive discretization method is proposed. The method based on the delay characteristics of pipelines is proposed to select temporal and spatial steps for reasonable approximation of DHN, and to achieve satisfactory accuracy results with a lower computational burden. Secondly, a two-stage OEF model for HE-IES with new linear approximations is developed. In this model, a detailed converter station model coupled AC grid and DC grid is investigated, and a linearization constraint method is proposed to handle the highly nonlinear of converter station by adding extended branches to the AC grid. Finally, simulations demonstrate that the effectiveness of the proposed model, which can provide more intact information of state variables for optimal planning.Note to Practitioners—This work is motivated by the demand for optimal energy management in heat and electricity integrated energy system (HE-IES). On the one hand, the slow dynamic characteristic of the district heating network (DHN) is crucial for calculating the optimal results. On the other hand, the improvement of electronic technology has diversified energy supplies and enriched the way of using electricity, such as converter-based generators for electric power grid (EPG) and converter-based boilers for DHN. Therefore, the model of HE-IES exhibits highly nonlinearity, which brings huge challenges to optimal energy flow analysis, especially for the real-time OEF analysis with state information updated every 15–30 min. Existing studies focus on computing optimal energy flow without considering the slow dynamic characteristic of DHN or the hybrid AC/DC scenarios brought by the converter-based sources. This paper presents an optimal energy flow problem that integrates the slow dynamic characteristic and the converter-based sources to retain the complete state information of the system as much as possible in the optimization process. The work is used on an integrated IEEE-33 bus and Denmark-61 node system to show the effectiveness.

Adversarial purification refers to a class of defense methods that remove adversarial perturbations using a generative model. These methods do not make assumptions on the form of attack and the classification model, and thus can defend pre-existing classifiers against unseen threats. However, their performance currently falls behind adversarial training methods. In this work, we propose DiffPure that uses diffusion models for adversarial purification: Given an adversarial example, we first diffuse it with a small amount of noise following a forward diffusion process, and then recover the clean image through a reverse generative process. To evaluate our method against strong adaptive attacks in an efficient and scalable way, we propose to use the adjoint method to compute full gradients of the reverse generative process. Extensive experiments on three image datasets including CIFAR-10, ImageNet and CelebA-HQ with three classifier architectures including ResNet, WideResNet and ViT demonstrate that our method achieves the state-of-the-art results, outperforming current adversarial training and adversarial purification methods, often by a large margin.

Certified robustness is a desirable property for deep neural networks in safety-critical applications, and popular training algorithms can certify robustness of a neural network by computing a global bound on its Lipschitz constant. However, such a bound is often loose: it tends to over-regularize the neural network and degrade its natural accuracy. A tighter Lipschitz bound may provide a better tradeoff between natural and certified accuracy, but is generally hard to compute exactly due to non-convexity of the network. In this work, we propose an efficient and trainable \emph{local} Lipschitz upper bound by considering the interactions between activation functions (e.g. ReLU) and weight matrices. Specifically, when computing the induced norm of a weight matrix, we eliminate the corresponding rows and columns where the activation function is guaranteed to be a constant in the neighborhood of each given data point, which provides a provably tighter bound than the global Lipschitz constant of the neural network. Our method can be used as a plug-in module to tighten the Lipschitz bound in many certifiable training algorithms. Furthermore, we propose to clip activation functions (e.g., ReLU and MaxMin) with a learnable upper threshold and a sparsity loss to assist the network to achieve an even tighter local Lipschitz bound. Experimentally, we show that our method consistently outperforms state-of-the-art methods in both clean and certified accuracy on MNIST, CIFAR-10 and TinyImageNet datasets with various network architectures.

Neural networks are vulnerable to input perturbations such as additive noise and adversarial attacks. In contrast, human perception is much more robust to such perturbations. The Bayesian brain hypothesis states that human brains use an internal generative model to update the posterior beliefs of the sensory input. This mechanism can be interpreted as a form of self-consistency between the maximum a posteriori (MAP) estimation of an internal generative model and the external environment. Inspired by such hypothesis, we enforce self-consistency in neural networks by incorporating generative recurrent feedback. We instantiate this design on convolutional neural networks (CNNs). The proposed framework, termed Convolutional Neural Networks with Feedback (CNN-F), introduces a generative feedback with latent variables to existing CNN architectures, where consistent predictions are made through alternating MAP inference under a Bayesian framework. In the experiments, CNN-F shows considerably improved adversarial robustness over conventional feedforward CNNs on standard benchmarks.

Pedestrian misalignment, which mainly arises from detector errors and pose variations, is a critical problem for a robust person re-identification (re-ID) system. With poor alignment, the feature learning and matching process might be largely compromised. To address this problem, this paper introduces the pose invariant embedding (PIE) as a pedestrian descriptor. First, in order to align pedestrians to a standard pose, the PoseBox structure is introduced, which is generated through pose estimation followed by affine transformations. Second, to reduce the impact of pose estimation errors and information loss during PoseBox construction, we design a PoseBox fusion (PBF) CNN architecture that takes the original image, the PoseBox, and the pose estimation confidence as input. The proposed PIE descriptor is thus defined as the fully connected layer of the PBF network for the retrieval task. Experiments are conducted on the Market-1501, CUHK03-NP, and DukeMTMC-reID datasets. We show that PoseBox alone yields decent re-ID accuracy, and that when integrated in the PBF network, the learned PIE descriptor produces competitive performance compared with the state-of-the-art approaches.

We present a system for converting a fully panoramic (360 degree) video into a normal field-of-view (NFOV) hyperlapse for an optimal viewing experience. Our system exploits visual saliency and semantics to non-uniformly sample in space and time for generating hyperlapses. In addition, users can optionally choose objects of interest for customizing the hyperlapses. We first stabilize an input 360 degree video by smoothing the rotation between adjacent frames and then compute regions of interest and saliency scores. An initial hyperlapse is generated by optimizing the saliency and motion smoothness followed by the saliency-aware frame selection. We further smooth the result using an efficient 2D video stabilization approach that adaptively selects the motion model to generate the final hyperlapse. We validate the design of our system by showing results for a variety of scenes and comparing against the state-of-the-art method through a large-scale user study.