Egocentric point tracking aims to localize points on object surfaces from a first-person perspective and serves as a critical step toward embodied intelligence. Recent methods rely on video input, tracking query points through feature matching across consecutive frames. However, these methods struggle in highly dynamic settings—a common challenge in first-person perspectives, where the head-mounted camera undergoes frequent and abrupt rotations, resulting in high angular velocities, motion blur, and large inter-frame displacements. In contrast, event cameras capture motion at microsecond temporal resolution, naturally avoiding blur and delivering low-latency, high-fidelity cues crucial for egocentric point tracking. Moreover, rapid egocentric motion disrupts local smoothness, breaking the assumption that spatially adjacent regions share similar motion. Event dynamics expose global motion trends, guiding coherent modeling and consistent feature flow. Therefore, this paper proposes a mamba-based tracking framework that constructs feature modeling paths aligned with the dominant motion trend extracted from events, and modulates feature propagation along these paths based on local motion intensity, enhancing stability by suppressing unreliable signals and emphasizing consistent cues. Additionally, a motion-adaptive suppression module enhances temporal robustness by adaptively suppressing correlation features based on motion intensity variations, mitigating the effects of intensity fluctuations and partial observability. To facilitate research in this domain, a multimodal dataset named DVS-EgoPoints with both events and videos for egocentric point tracking is collected. Experiments on the DVS-EgoPoints dataset and a simulation benchmark demonstrate superior performance over state-of-the-art methods, especially under challenging motion and occlusion conditions.

Tropical cyclone (TC) intensity forecasting is crucial for early disaster warning and emergency decision-making. Numerous researchers have explored deep-learning methods to address computational and post-processing issues in operational forecasting. Regrettably, they exhibit subpar long-term forecasting capabilities. We use two strategies to enhance long-term forecasting. (1) By enhancing the matching between TC intensity and spatial information, we can improve long-term forecasting performance. (2) Incorporating physical knowledge and physical constraints can help mitigate the accumulation of forecasting errors. To achieve the above strategies, we propose the VQLTI framework. VQLTI transfers the TC intensity information to a discrete latent space while retaining the spatial information differences, using large-scale spatial meteorological data as conditions. Furthermore, we leverage the forecast from the weather prediction model FengWu to provide additional physical knowledge for VQLTI. Additionally, we calculate the potential intensity (PI) to impose physical constraints on the latent variables. In the global long-term TC intensity forecasting, VQLTI achieves state-of-the-art results for the 24h to 120h, with the MSW (Maximum Sustained Wind) forecast error reduced by 35.65%-42.51% compared to ECMWF-IFS.

Accurate forecasting of tropical cyclone (TC) intensity is crucial for formulating disaster risk reduction strategies. Current methods predominantly rely on limited spatiotemporal information from ERA5 data and neglect the causal relationships between these physical variables, failing to fully capture the spatial and temporal patterns required for intensity forecasting. To address this issue, we propose a Multi-modal multi-Scale Causal AutoRegressive model (MSCAR), which is the first model that combines causal relationships with large-scale multimodal data for global TC intensity autoregressive forecasting. Furthermore, given the current absence of a TC dataset that offers a wide range of spatial variables, we present the Satellite and ERA5-based Tropical Cyclone Dataset (SETCD), which stands as the longest and most comprehensive global dataset related to TCs. Experiments on the dataset show that MSCAR outperforms the state-of-the-art methods, achieving maximum reductions in global and regional forecast errors of 9.52% and 6.74%, respectively. The code and dataset are publicly available at https://github.com/1457756434/MSCAR.git.

Machine learning (ML) techniques have shown great potential in generating high-quality solutions for integer linear programs (ILPs). However, existing methods typically rely on a *supervised learning* paradigm, leading to (1) *expensive training cost* due to repeated invocations of traditional solvers to generate training labels, and (2) *plausible yet infeasible solutions* due to the misalignment between the training objective (minimizing prediction loss) and the inference objective (generating high-quality solutions). To tackle this challenge, we propose **DiffILO** (**Diff**erentiable **I**nteger **L**inear Programming **O**ptimization), an *unsupervised learning paradigm for learning to solve ILPs*. Specifically, through a novel probabilistic modeling, DiffILO reformulates ILPs---discrete and constrained optimization problems---into continuous, differentiable (almost everywhere), and unconstrained optimization problems. This reformulation enables DiffILO to simultaneously solve ILPs and train the model via straightforward gradient descent, providing two major advantages. First, it significantly reduces the training cost, as the training process does not need the aid of traditional solvers at all. Second, it facilitates the generation of feasible and high-quality solutions, as the model *learns to solve ILPs* in an end-to-end manner, thus aligning the training and inference objectives. Experiments on commonly used ILP datasets demonstrate that DiffILO not only achieves an average training speedup of $13.2$ times compared to supervised methods, but also outperforms them by generating heuristic solutions with significantly higher feasibility ratios and much better solution qualities.

Message passing-based graph neural networks (GNNs) have achieved great success in many real-world applications. For a sampled mini-batch of target nodes, the message passing process is divided into two parts: message passing between nodes within the batch (MP-IB) and message passing from nodes outside the batch to those within it (MP-OB). However, MP-OB recursively relies on higher-order out-of-batch neighbors, leading to an exponentially growing computational cost with respect to the number of layers. Due to the neighbor explosion, the whole message passing stores most nodes and edges on the GPU such that many GNNs are infeasible to large-scale graphs. To address this challenge, we propose an accurate and fast mini-batch approach for large graph transductive learning, namely topological compensation (TOP), which obtains the outputs of the whole message passing solely through MP-IB, without the costly MP-OB. The major pillar of TOP is a novel concept of message invariance, which defines message-invariant transformations to convert costly MP-OB into fast MP-IB. This ensures that the modified MP-IB has the same output as the whole message passing. Experiments demonstrate that TOP is significantly faster than existing mini-batch methods by order of magnitude on vast graphs (millions of nodes and billions of edges) with limited accuracy degradation.

Generalization in visual reinforcement learning (VRL) aims to learn agents that can adapt to test environments with unseen visual distractions. Despite advances in robust representations learning, many methods do not take into account the essential downstream task of sequential decision-making. This leads to representations that lack critical long-term information, impairing decision-making abilities in test environments. To tackle this problem, we propose a novel robust action-value representation learning (ROUSER) under the information bottleneck (IB) framework. ROUSER learns robust representations to capture long-term information from the decision-making objective (i.e., action values). Specifically, ROUSER uses IB to encode robust representations by maximizing their mutual information with action values for long-term information, while minimizing mutual information with state-action pairs to discard irrelevant features. As action values are unknown, ROUSER proposes to decompose robust representations of state-action pairs into one-step rewards and robust representations of subsequent pairs. Thus, it can use known rewards to compute the loss for robust representation learning. Moreover, we show that ROUSER accurately estimates action values using learned robust representations, making it applicable to various VRL algorithms. Experiments demonstrate that ROUSER outperforms several state-of-the-art methods in eleven out of twelve tasks, across both unseen background and color distractions.

Chip placement is a critical step in the Electronic Design Automation (EDA) workflow, which aims to arrange chip modules on the canvas to optimize the performance, power, and area (PPA) metrics of final designs.Recent advances show great potential of AI-based algorithms in chip placement.However, due to the lengthy EDA workflow, evaluations of these algorithms often focus on intermediate surrogate metrics, which are computationally efficient but often misalign with the final end-to-end performance (i.e., the final design PPA).To address this challenge, we propose to build ChiPBench, a comprehensive benchmark specifically designed to evaluate the effectiveness of AI-based algorithms in final design PPA metrics.Specifically, we generate a diverse evaluation dataset from $20$ circuits across various domains, such as CPUs, GPUs, and NPUs.We then evaluate six state-of-the-art AI-based chip placement algorithms on the dataset and conduct a thorough analysis of their placement behavior.Extensive experiments show that AI-based chip placement algorithms produce unsatisfactory final PPA results, highlighting the significant influence of often-overlooked factors like regularity and dataflow.We believe ChiPBench will effectively bridge the gap between academia and industry.

With the development of large language models (LLMs), efficient inference through Key-Value (KV) cache compression has attracted considerable attention, especially for long-context generation. To compress the KV cache, recent methods identify critical KV tokens through static modeling of attention scores. However, these methods often struggle to accurately determine critical tokens as they neglect the *temporal patterns* in attention scores, resulting in a noticeable degradation in LLM performance. To address this challenge, we propose **AttentionPredictor**, which is the **first learning-based method to directly predict attention patterns for KV cache compression and critical token identification**. Specifically, AttentionPredictor learns a lightweight, unified convolution model to dynamically capture spatiotemporal patterns and predict the next-token attention scores. An appealing feature of AttentionPredictor is that it accurately predicts the attention score and shares the unified prediction model, which consumes negligible memory, among all transformer layers. Moreover, we propose a cross-token critical cache prefetching framework that hides the token estimation time overhead to accelerate the decoding stage. By retaining most of the attention information, AttentionPredictor achieves **13$\times$** KV cache compression and **5.6$\times$** speedup in a cache offloading scenario with comparable LLM performance, significantly outperforming the state-of-the-arts. The code is available at https://github.com/MIRALab-USTC/LLM-AttentionPredictor.

Few-shot image generation aims to train generative models using a small number of training images. When there are few images available for training (e.g. 10 images), Learning From Scratch (LFS) methods often generate images that closely resemble the training data while Transfer Learning (TL) methods try to improve performance by leveraging prior knowledge from GANs pre-trained on large-scale datasets. However, current TL methods may not allow for sufficient control over the degree of knowledge preservation from the source model, making them unsuitable for setups where the source and target domains are not closely related. To address this, we propose a novel pipeline called Peer is your Pillar (PIP), which combines a target few-shot dataset with a peer dataset to create a data-unbalanced conditional generation. Our approach includes a class embedding method that separates the class space from the latent space, and we use a direction loss based on pre-trained CLIP to improve image diversity. Experiments on various few-shot datasets demonstrate the advancement of the proposed PIP, especially reduces the training requirements of few-shot image generation.

Deep Neural Networks (DNNs) have achieved remarkable success across a wide range of tasks; however, their susceptibility to backdoor attacks remains a significant concern. Existing methods predominantly focus on optimizing the construction phase of backdoor attacks, aiming to reduce the detectability of trigger patterns and enhance stealth. In contrast, the selection phase—specifically the identification of appropriate benign samples for poisoning—has received limited attention. Recent studies have explored efficient poisoning sample selection to improve attack stealth. However, the underlying factors that determine the informativeness or effectiveness of a sample for backdoor learning remain poorly understood. To address this gap, we investigate the role of forgettable event and loss landscape curvature in enhancing poisoning sample efficiency. Our findings reveal that samples most likely to be forgotten during the poisoning process are crucial for effective attacks, and that low-curvature regions of the loss surface correlate with higher poisoning efficiency. Based on these insights, we introduce the Improved Filtering and Updating Strategy (FUS++), which significantly outperforms traditional selection methods in terms of efficiency. Our contributions provide new perspectives on sample selection for backdoor attacks and propose a novel strategy to improve poisoning efficacy.

The exponential growth of large language models (LLMs) has opened up numerous possibilities for multi-modal AGI systems. However, the progress in vision and vision-language foundation models, which are also critical elements of multi-modal AGI, has not kept pace with LLMs. In this work, we design a large-scale vision-language foun-dation model (Intern VL), which scales up the vision foun-dation model to 6 billion parameters and progressively aligns it with the LLM, using web-scale image-text data from various sources. This model can be broadly applied to and achieve state-of-the-art performance on 32 generic visual-linguistic benchmarks including visual perception tasks such as image-level or pixel-level recognition, vision-language tasks such as zero-shot image/video classification, zero-shot image/video-text retrieval, and link with LLMs to create multi-modal dialogue systems. It has powerful visual capabilities and can be a good alternative to the ViT-22B. We hope that our research could contribute to the development of multi-modal large models.

Machine learning (ML) has been shown to successfully accelerate solving NP-hard combinatorial optimization (CO) problems under the branch and bound framework. However, the high training and inference cost and limited interpretability of ML approaches severely limit their wide application to modern exact CO solvers. In contrast, human-designed policies---though widely integrated in modern CO solvers due to their compactness and reliability---can not capture data-driven patterns for higher performance. To combine the advantages of the two paradigms, we propose the first symbolic discovery framework---namely, deep symbolic discovery for exact combinatorial optimization solver (Symb4CO)---to learn high-performance symbolic policies on the branching task. Specifically, we show the potential existence of small symbolic policies empirically, employ a large neural network to search in the high-dimensional discrete space, and compile the learned symbolic policies directly for fast deployment. Experiments show that the Symb4CO learned purely CPU-based policies consistently achieve *comparable* performance to previous GPU-based state-of-the-art approaches. Furthermore, the appealing features of Symb4CO include its high training (*ten training instances*) and inference (*one CPU core*) efficiency and good interpretability (*one-line expressions*), making it simple and reliable for deployment. The results show encouraging potential for the *wide* deployment of ML to modern CO solvers.

Recent deep neural networks (DNNs) have came to rely on vast amounts of training data, providing an opportunity for malicious attackers to exploit and contaminate the data to carry out backdoor attacks. However, existing backdoor attack methods make unrealistic assumptions, assuming that all training data comes from a single source and that attackers have full access to the training data. In this paper, we introduce a more realistic attack scenario where victims collect data from multiple sources, and attackers cannot access the complete training data. We refer to this scenario as $\textbf{data-constrained backdoor attacks}$. In such cases, previous attack methods suffer from severe efficiency degradation due to the $\textbf{entanglement}$ between benign and poisoning features during the backdoor injection process. To tackle this problem, we introduce three CLIP-based technologies from two distinct streams: $\textit{Clean Feature Suppression}$ and $\textit{Poisoning Feature Augmentation}$. The results demonstrate remarkable improvements, with some settings achieving over $\textbf{100}$% improvement compared to existing attacks in data-constrained scenarios.

Learning neural operators for solving partial differential equations (PDEs) has attracted great attention due to its high inference efficiency. However, training such operators requires generating a substantial amount of labeled data, i.e., PDE problems together with their solutions. The data generation process is exceptionally time-consuming, as it involves solving numerous systems of linear equations to obtain numerical solutions to the PDEs. Many existing methods solve these systems independently without considering their inherent similarities, resulting in extremely redundant computations. To tackle this problem, we propose a novel method, namely **S**orting **K**rylov **R**ecycling (**SKR**), to boost the efficiency of solving these systems, thus significantly accelerating data generation for neural operators training. To the best of our knowledge, SKR is the first attempt to address the time-consuming nature of data generation for learning neural operators. The working horse of SKR is Krylov subspace recycling, a powerful technique for solving a series of interrelated systems by leveraging their inherent similarities. Specifically, SKR employs a sorting algorithm to arrange these systems in a sequence, where adjacent systems exhibit high similarities. Then it equips a solver with Krylov subspace recycling to solve the systems sequentially instead of independently, thus effectively enhancing the solving efficiency. Both theoretical analysis and extensive experiments demonstrate that SKR can significantly accelerate neural operator data generation, achieving a remarkable speedup of up to 13.9 times.

Generation of plausible but incorrect factual information, often termed hallucination, has attracted significant research interest. Retrieval-augmented language model (RALM)---which enhances models with up-to-date knowledge---emerges as a promising method to reduce hallucination. However, existing RALMs may instead exacerbate hallucination when retrieving lengthy contexts. To address this challenge, we propose COFT, a novel **CO**arse-to-**F**ine highligh**T**ing method to focus on different granularity-level key texts, thereby avoiding getting lost in lengthy contexts. Specifically, COFT consists of three components: *recaller*, *scorer*, and *selector*. First, *recaller* applies a knowledge graph to extract potential key entities in a given context. Second, *scorer* measures the importance of each entity by calculating its contextual weight. Finally, *selector* selects high contextual weight entities with a dynamic threshold algorithm and highlights the corresponding paragraphs, sentences, or words in a coarse-to-fine manner. Extensive experiments on knowledge hallucination benchmark demonstrate the effectiveness of COFT, leading to a superior performance over 30% in F1 score metric. Moreover, COFT also exhibits remarkable versatility across various long-form tasks, such as reading comprehension and question answering.

Mixed-integer linear programming (MILP) is one of the most popular mathematical formulations with numerous applications. In practice, improving the performance of MILP solvers often requires a large amount of high-quality data, which can be challenging to collect. Researchers thus turn to generation techniques to generate additional MILP instances. However, existing approaches do not take into account specific block structures—which are closely related to the problem formulations—in the constraint coefficient matrices (CCMs) of MILPs. Consequently, they are prone to generate computationally trivial or infeasible instances due to the disruptions of block structures and thus problem formulations. To address this challenge, we propose a novel MILP generation framework, called Block Structure Decomposition (MILP-StuDio), to generate high-quality instances by preserving the block structures. Specifically, MILP-StuDio begins by identifying the blocks in CCMs and decomposing the instances into block units, which serve as the building blocks of MILP instances. We then design three operators to construct new instances by removing, substituting, and appending block units in the original instances, enabling us to generate instances with flexible sizes. An appealing feature of MILP-StuDio is its strong ability to preserve the feasibility and computational hardness of the generated instances. Experiments on the commonly-used benchmarks demonstrate that using instances generated by MILP-StuDio is able to significantly reduce over 10% of the solving time for learning-based solvers.

Logic Synthesis (LS) aims to generate an optimized logic circuit satisfying a given functionality, which generally consists of circuit translation and optimization. It is a challenging and fundamental combinatorial optimization problem in integrated circuit design. Traditional LS approaches rely on manually designed heuristics to tackle the LS task, while machine learning recently offers a promising approach towards next-generation logic synthesis by neural circuit generation and optimization. In this paper, we first revisit the application of differentiable neural architecture search (DNAS) methods to circuit generation and found from extensive experiments that existing DNAS methods struggle to exactly generate circuits, scale poorly to large circuits, and exhibit high sensitivity to hyper-parameters. Then we provide three major insights for these challenges from extensive empirical analysis: 1) DNAS tends to overfit to too many skip-connections, consequently wasting a significant portion of the network's expressive capabilities; 2) DNAS suffers from the structure bias between the network architecture and the circuit inherent structure, leading to inefficient search; 3) the learning difficulty of different input-output examples varies significantly, leading to severely imbalanced learning. To address these challenges in a systematic way, we propose a novel regularized triangle-shaped circuit network generation framework, which leverages our key insights for completely accurate and scalable circuit generation. Furthermore, we propose an evolutionary algorithm assisted by reinforcement learning agent restarting technique for efficient and effective neural circuit optimization. Extensive experiments on four different circuit benchmarks demonstrate that our method can precisely generate circuits with up to 1200 nodes. Moreover, our synthesized circuits significantly outperform the state-of-the-art results from several competitive winners in IWLS 2022 and 2023 competitions.

Recently, vision model pre-training has evolved from relying on manually annotated datasets to leveraging large-scale, web-crawled image-text data. Despite these advances, there is no pre-training method that effectively exploits the interleaved image-text data, which is very prevalent on the Internet. Inspired by the recent success of compression learning in natural language processing, we propose a novel vision model pre-training method called Latent Compression Learning (LCL) for interleaved image-text data. This method performs latent compression learning by maximizing the mutual information between the inputs and outputs of a causal attention model. The training objective can be decomposed into two basic tasks: 1) contrastive learning between visual representation and preceding context, and 2) generating subsequent text based on visual representation. Our experiments demonstrate that our method not only matches the performance of CLIP on paired pre-training datasets (e.g., LAION), but can also leverage interleaved pre-training data (e.g., MMC4) to learn robust visual representations from scratch, showcasing the potential of vision model pre-training with interleaved image-text data.

Poisoning efficiency is crucial in poisoning-based backdoor attacks, as attackers aim to minimize the number of poisoning samples while maximizing attack efficacy. Recent studies have sought to enhance poisoning efficiency by selecting effective samples. However, these studies typically rely on a proxy backdoor injection task to identify an efficient set of poisoning samples. This proxy attack-based approach can lead to performance degradation if the proxy attack settings differ from those of the actual victims, due to the shortcut nature of backdoor learning. Furthermore, proxy attack-based methods are extremely time-consuming, as they require numerous complete backdoor injection processes for sample selection. To address these concerns, we present a Proxy attack-Free Strategy (PFS) designed to identify efficient poisoning samples based on the similarity between clean samples and their corresponding poisoning samples, as well as the diversity of the poisoning set. The proposed PFS is motivated by the observation that selecting samples with high similarity between clean and corresponding poisoning samples results in significantly higher attack success rates compared to using samples with low similarity. Additionally, we provide theoretical foundations to explain the proposed PFS. We comprehensively evaluate the proposed strategy across various datasets, triggers, poisoning rates, architectures, and training hyperparameters. Our experimental results demonstrate that PFS enhances backdoor attack efficiency while also offering a remarkable speed advantage over previous proxy attack-based selection methodologies.

To effectively exploit the potential of large-scale models, various pre-training strategies supported by massive data from different sources are proposed, including supervised pre-training, weakly-supervised pre-training, and self-supervised pre-training. It has been proved that combining multiple pre-training strategies and data from various modalities/sources can greatly boost the training of large-scale models. However, current works adopt a multi-stage pre-training system, where the complex pipeline may increase the uncertainty and instability of the pre-training. It is thus desirable that these strategies can be integrated in a single-stage manner. In this paper, we first propose a general multimodal mutual information formula as a unified optimization target and demonstrate that all mainstream approaches are special cases of our framework. Under this unified perspective, we propose an all-in-one single-stage pre-training approach, named Maximizing Multi-modal Mutual Information Pre-Training (M3I Pre-training). Our approach achieves better performance than previous pre-training methods on various vision benchmarks, including ImageNet classification, COCO object detection, LVIS long-tailed object detection, and ADE20k semantic segmentation. Notably, we successfully pre-train a billion-level parameter image backbone and achieve state-of-the-art performance on various benchmarks under public data setting. Code shall be released at https://github.com/OpenGVLab/M3I-Pre-Training.

Self-supervised learning (SSL) has delivered superior performance on a variety of downstream vision tasks. Two main-stream SSL frameworks have been proposed, i.e., Instance Discrimination (ID) and Masked Image Modeling (MIM). ID pulls together representations from different views of the same image, while avoiding feature collapse. It lacks spatial sensitivity, which requires modeling the local structure within each image. On the other hand, MIM reconstructs the original content given a masked image. It instead does not have good semantic alignment, which requires projecting semantically similar views into nearby representations. To address this dilemma, we observe that (1) semantic alignment can be achieved by matching different image views with strong augmentations; (2) spatial sensitivity can benefit from predicting dense representations with masked images. Driven by these analysis, we propose Siamese Image Modeling (SiameseIM), which predicts the dense representations of an augmented view, based on another masked view from the same image but with different augmentations. SiameseIM uses a Siamese network with two branches. The online branch encodes the first view, and predicts the second view's representation according to the relative positions between these two views. The target branch produces the target by encoding the second view. SiameseIM can surpass both ID and MIM on a wide range of downstream tasks, including ImageNet finetuning and linear probing, COCO and LVIS detection, and ADE20k semantic segmentation. The improvement is more significant in few-shot, long-tail and robustness-concerned scenarios. Code shall be released.

Transparent objects are widely used in industrial automation and daily life. However, robust visual recognition and perception of transparent objects have always been a major challenge. Currently, most commercial-grade depth cameras are still not good at sensing the surfaces of transparent objects due to the refraction and reflection of light. In this work, we present a transformer-based transparent object depth estimation approach from a single RGB-D input. We observe that the global characteristics of the transformer make it easier to extract contextual information to perform depth estimation of transparent areas. In addition, to better enhance the fine-grained features, a feature fusion module (FFM) is designed to assist coherent prediction. Our empirical evidence demonstrates that our model delivers significant improvements in recent popular datasets, e.g., 25% gain on RMSE and 21% gain on REL compared to previous state-of-the-art convolutional-based counterparts in ClearGrasp dataset. Extensive results show that our transformer-based model enables better aggregation of the object's RGB and inaccurate depth information to obtain a better depth representation. Our code and the pre-trained model are available at https://github.com/yuchendoudou/TODE.

In this study, we delve into the task of few-shot Generative Domain Adaptation (GDA), which involves transferring a pre-trained generator from one domain to a new domain using only a few reference images. Inspired by the way human brains acquire knowledge in new domains, we present an innovative generator structure called $\textbf{Domain Re-Modulation (DoRM)}$. DoRM not only meets the criteria of $\textit{high quality}$, $\textit{large synthesis diversity}$, and $\textit{cross-domain consistency}$, which were achieved by previous research in GDA, but also incorporates $\textit{memory}$ and $\textit{domain association}$, akin to how human brains operate. Specifically, DoRM freezes the source generator and introduces new mapping and affine modules (M\&A modules) to capture the attributes of the target domain during GDA. This process resembles the formation of new synapses in human brains. Consequently, a linearly combinable domain shift occurs in the style space. By incorporating multiple new M\&A modules, the generator gains the capability to perform high-fidelity multi-domain and hybrid-domain generation. Moreover, to maintain cross-domain consistency more effectively, we introduce a similarity-based structure loss. This loss aligns the auto-correlation map of the target image with its corresponding auto-correlation map of the source image during training. Through extensive experiments, we demonstrate the superior performance of our DoRM and similarity-based structure loss in few-shot GDA, both quantitatively and qualitatively. Code will be available at https://github.com/wuyi2020/DoRM.

Counterfactual regret minimization (CFR) has achieved many fascinating results in solving large-scale imperfect information games (IIGs). Neural network approximation CFR (neural CFR) is one of the promising techniques that can reduce computation and memory consumption by generalizing decision information between similar states. Current neural CFR algorithms have to approximate cumulative regrets. However, efficient and accurate approximation in a large-scale IIG is still a tough challenge. In this article, a new CFR variant, recursive CFR (ReCFR), is proposed. In ReCFR, recursive substitute values (RSVs) are learned and used to replace cumulative regrets. It is proven that ReCFR can converge to a Nash equilibrium at a rate of$O({1}/{\sqrt{T}})$. Based on ReCFR, a new model-free neural CFR with bootstrap learning, neural ReCFR-B, is proposed. Due to the recursive and noncumulative nature of RSVs, neural ReCFR-B has lower variance training targets than other neural CFRs. Experimental results show that neural ReCFR-B is competitive with the state-of-the-art neural CFR algorithms at a much lower training cost.

While Deep Neural Networks (DNNs) excel in many tasks, the huge training resources they require become an obstacle for practitioners to develop their own models. It has become common to collect data from the Internet or hire a third party to train models. Unfortunately, recent studies have shown that these operations provide a viable pathway for maliciously injecting hidden backdoors into DNNs. Several defense methods have been developed to detect malicious samples, with the common assumption that the latent representations of benign and malicious samples extracted by the infected model exhibit different distributions. However, it is still an open question whether this assumption holds up. In this article, we investigate such differences thoroughly via answering three questions: 1) What are the characteristics of the distributional differences? 2) How can they be effectively reduced? 3) What impact does this reduction have on difference-based defense methods? First, the distributional differences of multi-level representations on the regularly trained backdoored models are verified to be significant by adopting Maximum Mean Discrepancy (MMD), Energy Distance (ED), and Sliced Wasserstein Distance (SWD) as the metrics. Then, ML-MMDR, a difference reduction method that adds multi-level MMD regularization into the loss, is proposed, and its effectiveness is testified on three typical difference-based defense methods. Across all the experimental settings, the F1 scores of these methods drop from 90%-100% on the regularly trained backdoored models to 60%-70% on the models trained with ML-MMDR. These results indicate that the proposed MMD regularization can enhance the stealthiness of existing backdoor attack methods. The prototype code of our method is now available athttps://github.com/xpf/Multi-Level-MMD-Regularization.

Generative Adversarial Networks (GANs) have the ability to generate images that are visually indistinguishable from real images. However, recent studies have revealed that generated and real images share significant differences in the frequency domain. In this article, we argue that the frequency gap is caused by the high-frequency sensitivity of the discriminator. According to our observation, during the training of most GANs, severe high-frequency differences make the discriminator focus on high-frequency components excessively, which hinders the generator from fitting the low-frequency components that are important for learning images’ content. Then, we propose two simple yet effective image pre-processing operations in the frequency domain for eliminating the side effects caused by high-frequency differences in GANs training: High-frequency Confusion (HFC) and High-frequency Filter (HFF). The proposed operations are general and can be applied to most existing GANs at a fraction of the cost. The advanced performance of the proposed operations is verified on multiple loss functions, network architectures, and datasets. Specifically, the proposed HFF achieves significant improvements of 42.5% FID on CelebA (128*128) unconditional generation based on SNGAN, 30.2% FID on CelebA unconditional generation based on SSGAN, and 69.3% FID on CelebA unconditional generation based on InfoMAXGAN. Furthermore, we also adopt HFF as the first attempt at data augmentation in the frequency domain for contrastive learning, achieving state-of-the-art performance on unconditional generation. Code is available at https://github.com/iceli1007/HFC-and-HFF .

The multi-objective testing resource allocation problem (MOTRAP) aims at seeking a good trade-off between system reliability, testing cost, and testing time, which is of significant importance to facilitate the testing planning. Yet most studies focus on the time constraint but rarely consider the practical reliability requirement. In this work, we address MOTRAP on an architecture-based model (ABM) with the personalized preference over reliability. More specifically, we first present a reliability-constrained MOTRAP model on the basis of ABM and illustrate how to use this model for real-world systems. Then, to leverage the problem's knowledge, we develop new lower and upper bounds on testing time invested in different components from both theoretical and algorithmic perspectives on the basis of the Lagrange multiplier and half-interval search. Importantly, these new derived bounds have strong implications due to the fact that they can be easily employed by optimizers as the limits of variables to prune the search space to the region of interests of the decision maker and locate feasible solutions with the expected reliability. Finally, we evaluate the proposed bounds in popular multi-objective optimizers for MOTRAP on application and empirical cases. Experimental results demonstrate that our new bounds practically improve the search performance of optimizers, and decision makers can easily combine these new bounds with off-the-shelf optimizers to find higher-quality solutions that they are interested in, which greatly soothes away stress on optimizer and solution selections of decision makers.

Deep reinforcement learning algorithms can perform poorly in real-world tasks due to the discrepancy between source and target environments. This discrepancy is commonly viewed as the disturbance in transition dynamics. Many existing algorithms learn robust policies by modeling the disturbance and applying it to source environments during training, which usually requires prior knowledge about the disturbance and control of simulators. However, these algorithms can fail in scenarios where the disturbance from target environments is unknown or is intractable to model in simulators. To tackle this problem, we propose a novel model-free actor-critic algorithm---namely, state-conservative policy optimization (SCPO)---to learn robust policies without modeling the disturbance in advance. Specifically, SCPO reduces the disturbance in transition dynamics to that in state space and then approximates it by a simple gradient-based regularizer. The appealing features of SCPO include that it is simple to implement and does not require additional knowledge about the disturbance or specially designed simulators. Experiments in several robot control tasks demonstrate that SCPO learns robust policies against the disturbance in transition dynamics.

Model-based reinforcement learning algorithms, which aim to learn a model of the environment to make decisions, are more sample efficient than their model-free counterparts. The sample efficiency of model-based approaches relies on whether the model can well approximate the environment. However, learning an accurate model is challenging, especially in complex and noisy environments. To tackle this problem, we propose the conservative model-based actor-critic (CMBAC), a novel approach that achieves high sample efficiency without the strong reliance on accurate learned models. Specifically, CMBAC learns multiple estimates of the Q-value function from a set of inaccurate models and uses the average of the bottom-k estimates---a conservative estimate---to optimize the policy. An appealing feature of CMBAC is that the conservative estimates effectively encourage the agent to avoid unreliable “promising actions”---whose values are high in only a small fraction of the models. Experiments demonstrate that CMBAC significantly outperforms state-of-the-art approaches in terms of sample efficiency on several challenging control tasks, and the proposed method is more robust than previous methods in noisy environments.

Feature Pyramid Network (FPN) has been a generic feature extractor in computer vision tasks, which utilizes multi-level features to generate discriminative pyramidal representations. However, the way simply using Sum or Concatenate operation on features to integrate multi-scale information is not sufficient to obtain discriminative semantic representations. In this paper, we propose a dynamic feature pyramid network (DyFPN) to merge multi-scale information in both features and weights. DyFPN uses both high-level context features and low-level spatial structural features to obtain dynamic convolution kernel that contains multi-scale information. In this manner, each resolution in the pyramid performs unique and adaptive convolution directly, meanwhile strengthening the information flow. Specially, DyFPN can be regarded as a complementary enhancement to existing feature pyramid networks. We analyze the effective receptive field and attention map of DyFPN. It proves that our method contains more local information and global information compared with merging multi-scale information only on feature level. Benefit from multi-ways of integrating multi-scale information, our method outperforms other existing feature pyramid methods on COCO detection tasks by a large margin.

Follow-the-Regularized-Leader (FTRL) and Online Mirror Descent (OMD) are regret minimization algorithms for Online Convex Optimization (OCO), they are mathematically elegant but less practical in solving Extensive-Form Games (EFGs). Counterfactual Regret Minimization (CFR) is a technique for approximating Nash equilibria in EFGs. CFR and its variants have a fast convergence rate in practice, but their theoretical results are not satisfactory. In recent years, researchers have been trying to link CFRs with OCO algorithms, which may provide new theoretical results and inspire new algorithms. However, existing analysis is restricted to local decision points. In this paper, we show that CFRs with Regret Matching and Regret Matching+ are equivalent to special cases of FTRL and OMD, respectively. According to these equivalences, a new FTRL and a new OMD algorithm, which can be considered as extensions of vanilla CFR and CFR+, are derived. The experimental results show that the two variants converge faster than conventional FTRL and OMD, even faster than vanilla CFR and CFR+ in some EFGs.

Recent studies have proven that deep neural networks are vulnerable to backdoor attacks. Specifically, by mixing a small number of poisoned samples into the training set, the behavior of the trained model can be maliciously controlled. Existing attack methods construct such adversaries by randomly selecting some clean data from the benign set and then embedding a trigger into them. However, this selection strategy ignores the fact that each poisoned sample contributes inequally to the backdoor injection, which reduces the efficiency of poisoning. In this paper, we formulate improving the poisoned data efficiency by the selection as an optimization problem and propose a Filtering-and-Updating Strategy (FUS) to solve it. The experimental results on CIFAR-10 and ImageNet-10 indicate that the proposed method is effective: the same attack success rate can be achieved with only 47% to 75% of the poisoned sample volume compared to the random selection strategy. More importantly, the adversaries selected according to one setting can generalize well to other settings, exhibiting strong transferability. The prototype code of our method is now available at https://github.com/xpf/Data-Efficient-Backdoor-Attacks.

Adversarial training (AT) commonly serves as an advanced regularization to establish enhanced robust models. However, it usually scarifies performance on clean inputs, especially in complicated object detection and semantic segmentation tasks. However, how to fully unleash the power of adversarial training regularization to improve the trade-off between standard performance and adversarial robustness of detection and segmentation models, has not been explored. In this paper, we present the Vertical and Horizontal Adversarial Training (VHAT) regularization on both input and intermediate features, which consists of two major components: i) Vertical Adversarial Training (VAT) by utilizing adversarial features with a wide range of attack strengths; ii) Horizontal Adversarial Training (HAT) by injecting layer-wise adversarial feature perturbations together with adversarial samples. Extensive experiment results demonstrate that VHAT achieves the standard performance and adversarial robustness double-win for Faster-RCNN on PASCAL VOC and DeepLabv3+ on PASCAL VOC and Cityscapes datasets, respectively. Comprehensive ablation studies and visualizations are provided to reveal the insights and working mechanisms.

Generalization across different environments with the same tasks is critical for successful applications of visual reinforcement learning (RL) in real scenarios. However, visual distractions---which are common in real scenes---from high-dimensional observations can be hurtful to the learned representations in visual RL, thus degrading the performance of generalization. To tackle this problem, we propose a novel approach, namely Characteristic Reward Sequence Prediction (CRESP), to extract the task-relevant information by learning reward sequence distributions (RSDs), as the reward signals are task-relevant in RL and invariant to visual distractions. Specifically, to effectively capture the task-relevant information via RSDs, CRESP introduces an auxiliary task---that is, predicting the characteristic functions of RSDs---to learn task-relevant representations, because we can well approximate the high-dimensional distributions by leveraging the corresponding characteristic functions. Experiments demonstrate that CRESP significantly improves the performance of generalization on unseen environments, outperforming several state-of-the-arts on DeepMind Control tasks with different visual distractions.

Deep learning models have been found with a tendency of relying on shortcuts, i.e., decision rules that perform well on standard benchmarks but fail when transferred to more challenging testing conditions. Such reliance may hinder deep learning models from learning other task-related features and seriously affect their performance and robustness. Although recent studies have shown some characteristics of shortcuts, there are few investigations on how to help the deep learning models to solve shortcut problems. This paper proposes a framework to address this issue by setting up roadblocks on shortcuts. Specifically, roadblocks are placed when the model is urged to learn to complete a gently modified task to ensure that the learned knowledge, including shortcuts, is insufficient the complete the task. Therefore, the model trained on the modified task will no longer over-rely on shortcuts. Extensive experiments demonstrate that the proposed framework significantly improves the training of networks on both synthetic and real-world datasets in terms of both classification accuracy and feature diversity. Moreover, the visualization results show that the mechanism behind the proposed our method is consistent with our expectations. In summary, our approach can effectively disable the shortcuts and thus learn more robust features.

The multiobjective testing resource allocation problem (MOTRAP) is how to efficiently allocate the finite testing time to various modules, with the aim of optimizing system reliability, testing cost, and testing time simultaneously. To deal with this problem, a common approach is to use multiobjective evolutionary algorithms (MOEAs) to seek a set of tradeoff solutions between the three objectives. However, such a tradeoff set may contain a substantial proportion of solutions with very low reliability level, which consume lots of computational resources but may be valueless to the software project manager. In this article, a MOTRAP model with a prespecified reliability is first proposed. Then, new lower bounds on the testing time invested in different modules are theoretically deduced from the necessary condition for the achievement of the given reliability, based on which an exact algorithm for determining the new lower bounds is presented. Moreover, several enhanced constraint-handling techniques (ECHTs) derived from the new bounds are successively developed to be combined with MOEAs to correct and reduce the constraint violation. Finally, the proposed ECHTs are evaluated in comparison with various state-of-the-art constraint-solving approaches. The comparative results demonstrate that the proposed ECHTs can work well with MOEAs, make the search focus on the feasible region of the prespecified reliability, and provide the software project manager with better and more diverse, satisfactory choices in test planning.

Generative adversarial nets (GANs) have been successfully applied in many fields like image generation, inpainting, super-resolution, and drug discovery, etc. By now, the inner process of GANs is far from being understood. To get a deeper insight into the intrinsic mechanism of GANs, in this paper, a method for interpreting the latent space of GANs by analyzing the correlation between latent variables and the corresponding semantic contents in generated images is proposed. Unlike previous methods that focus on dissecting models via feature visualization, the emphasis of this work is put on the variables in latent space, i.e. how the latent variables affect the quantitative analysis of generated results. Given a pre-trained GAN model with weights fixed, the latent variables are intervened to analyze their effect on the semantic content in generated images. A set of controlling latent variables can be derived for specific content generation, and the controllable semantic content manipulation is achieved. The proposed method is testified on the datasets Fashion-MNIST and UT Zappos50K, experiment results show its effectiveness.

Automatic medical image segmentation has wide applications for disease diagnosing. However, it is much more challenging than natural optical image segmentation due to the high-resolution of medical images and the corresponding huge computation cost. The sliding window is a commonly used technique for whole slide image (WSI) segmentation, however, for these methods based on the sliding window, the main drawback is lacking global contextual information for supervision. In this paper, we propose a dual-inputs attention network (denoted as DA-RefineNet) for WSI segmentation, where both local fine-grained information and global coarse information can be efficiently utilized. Sufficient comparative experiments are conducted to evaluate the effectiveness of the proposed method, the results prove that the proposed method can achieve better performance on WSI segmentation compared to methods relying on single-input.

The intensive consumption of resources by evolutionary algorithms makes it very time-consuming to search for network architectures. In this paper, We proposed a efficient evolution method for neural architecture search. Our method adopts the weight sharing strategy, in which a supernet is built to subsume all architectures, to speed up architecture evaluation. A universal choice strategy is designed to deal with the inaccurate evaluation caused by the methods that speeding up evaluation. Instead of searching for the best architecture, we search for the set of excellent architectures and derive the final architecture from derive the target architecture according to commonalities of these architectures. The proposed method achieved better results(2.40% test error rate on CIFAR-10 with 3.66M parameters) compared to other the-state-of-art method using less than 0.4 GPU days.

The performance of deep reinforcement learning methods prone to degenerate when applied to tasks requiring relatively longer horizon memory or with highly variable dynamics. In this paper, we utilize the probabilistic latent context variables motivated by recent Meta-RL materials, and propose the Latent Context based Soft Actor-Critic (LC-SAC) approach to address aforementioned issues. The latent context is capable to encode information about both the agent's previous behaviors and the dynamics of the current undergoing environment, which empirically believed to be beneficial for efficient policy optimization. Experiment results demonstrate that LC-SAC can achieve comparable performance with SAC on a collection of continuous control benchmarks and outperforms SAC in some particular tasks with above two characteristics. Moreover, we also introduce a simple but general procedure to integrate LC-SAC with diverse-quality demonstrations to enable efficient reuse of human prior knowledge, and finally achieve competitive performance with comparatively small number of interactions with environments.

The rate of transient faults has increased significantly as the technology scales up. The tolerance of transient faults has become an important issue in the system design. Dual modular redundancy (DMR) and triple modular redundancy (TMR) are two commonly used techniques that can achieve fault detection and masking through executing redundant tasks. As DMR and TMR have different time and cost overheads, we must carefully determine which one should be used for each task (i.e., task hardening) to achieve the optimal system design. Furthermore, for multi-core systems, the system-level design includes the allocation of cores for the tasks (i.e., task mapping) as well. This paper aims at task hardening and mapping simultaneously for independent tasks on multi-cores with heterogeneous performances, in order to minimize the maximum completion time of all tasks (i.e., makespan). We demonstrate that once task hardening is given, task mapping of independent tasks can be achieved by employing min–max-weight perfect matching with a polynomial time complexity. Besides, as there is a trade-off between cost and time performance, we propose a multi-objective memetic algorithm (MOMA)-based task hardening method to obtain a set of solutions with different numbers of cores (i.e., costs), so the designer can choose different solutions according to different requirements. The key idea of the MOMA is to incorporate problem-specific knowledge into the global search of evolutionary algorithms. Our experimental studies have demonstrated the effectiveness of the proposed method and have shown that by combining the results of MOMA and MOEA we can provide a designer with a highly accurate set of solutions within a reasonable amount of time.

Coalition formation has been a fundamental form of resource cooperation for achieving joint goals in multiagent systems. Most existing studies still focus on the traditional assumption that an agent has to contribute its resources to all the goals, even if the agent is not interested in the goal at all. In this article, a natural extension of the traditional coalitional resource games (CRGs) is studied from both theoretical and empirical perspectives, in which each agent has uncompromising, personalized preferences over goals. Specifically, a new CRGs model with agents’ strict preferences for goals is presented, in which an agent is willing to contribute its resources only to the goals that are in its own interest set. The computational complexity of the basic decision problems surrounding the successful coalition is reinvestigated. The results suggest that these problems in such a strict preference way are complex and intractable. To find the largest successful coalition for possible computation reduction or potential parallel processing, a flow-network–based exhaust algorithm, called FNetEA, is proposed to achieve the optimal solution. Then, to solve the problem more efficiently, a hybrid algorithm, named 2D-HA, is developed to find the approximately optimal solution on the basis of genetic algorithm, two-dimensional (2D) solution representation, and a heuristic for solution repairs. Through extensive experiments, the 2D-HA algorithm exhibits the prominent ability to provide reassurances that the optimal solution could be found within a reasonable period of time, even in a super-large-scale space.

Metal artifacts commonly appear in computed tomography (CT) images of the patient body with metal implants and can affect disease diagnosis. Known deep learning and traditional metal trace restoring methods did not effectively restore details and sinogram consistency information in X-ray CT sinograms, hence often causing considerable secondary artifacts in CT images. In this paper, we propose a new cross-domain metal trace restoring network which promotes sinogram consistency while reducing metal artifacts and recovering tissue details in CT images. Our new approach includes a cross-domain procedure that ensures information exchange between the image domain and the sinogram domain in order to help them promote and complement each other. Under this cross-domain structure, we develop a hierarchical analytic network (HAN) to recover fine details of metal trace, and utilize the perceptual loss to guide HAN to concentrate on the absorption of sinogram consistency information of metal trace. To allow our entire cross-domain network to be trained end-to-end efficiently and reduce the graphic memory usage and time cost, we propose effective and differentiable forward projection (FP) and filtered back-projection (FBP) layers based on FP and FBP algorithms. We use both simulated and clinical datasets in three different clinical scenarios to evaluate our proposed network's practicality and universality. Both quantitative and qualitative evaluation results show that our new network outperforms state-of-the-art metal artifact reduction methods. In addition, the elapsed time analysis shows that our proposed method meets the clinical time requirement.

Cooperative Co-evolution (CC) is a promising framework to scale up conventional evolutionary algorithms for large scale global optimization (LSGO) problems. However, how to group decision variables is still a problem while there is no prior knowledge about the dependence relationship between variables. In this paper, a new kind of CC algorithm called Soft Grouping Cooperative Co-evolution (SGCC) is proposed to tackle the problem. Instead of explicitly dividing variables into multiple groups, the algorithm softly assigns variables into multiple groups by controlling the degree of membership of variables to the groups. In this work, the degree of membership is controlled by a probability distribution function. The experimental investigation shows that Soft Grouping CC is better than the explicit grouping CC on partially separable and non-separable problems.

Rolling Horizon Evolutionary Algorithm (RHEA) is an online planning method for real-time game playing; its performance is closely related to the planning horizon and the search cost allowed. In this paper, we propose to learn a prior for RHEA in an offline manner by training a value network and a policy network. The value network is used to reduce the planning horizon by providing an estimation of future rewards, and the policy network is used to initialize the population, which helps to narrow down the search scope. The proposed algorithm, named prior-based RHEA (p-RHEA), trains policy and value networks by performing planning and learning iteratively. In the planning stage, the horizon-limited search is performed to improve the policies and collect training samples with the help of the learned networks. In the learning stage, the policy network and value network are trained with the collected samples to learn better prior knowledge. Experimental results on OpenAI MuJoCo tasks show that the performance of the proposed p- RHEA is significantly improved compared to that of RHEA.

Fitness landscape analysis (FLA) is an important approach for studying how hard problems are for metaheuristic algorithms to solve. Static FLA focuses on extracting the properties of a problem and does not consider any information about the optimization algorithms; thus, it is not adequate for indicating whether a particular algorithm is suitable for solving a problem. By contrast, dynamic FLA considers the behavior of algorithms in combination with the properties of an optimization problem to determine the effectiveness of a given algorithm for solving that problem. However, previous dynamic FLA approaches are all individually based and lack statistical significance. In this paper, the concept of population evolvability is presented, as an extension of dynamic FLA, to quantify the effectiveness of population-based metaheuristic algorithms for solving a given problem. Specifically, two measures of population evolvability are defined that describe the probability that a population will obtain improved solutions to a problem and its ability to do so. Then, a combined measure is derived from these two measures to represent the overall population evolvability. Subsequently, the significance and validity of the proposed measures are investigated through analytical and experimental studies. Finally, the utility of the proposed measures is illustrated in an application of algorithm selection for black-box optimization problems. High accuracy in selecting the best algorithm is observed in a statistical analysis, with a low computational cost in terms of fitness evaluations.

High defect density and extreme parameter variation make it very difficult to implement reliable logic functions in crossbar-based nanoarchitectures. It is a major design challenge to tolerate defects and variations simultaneously for such architectures. In this paper, a method based on a bipartite matching and memetic algorithm is proposed for defect- and variation-tolerant logic mapping (D/VTLM) problem in crossbar-based nanoarchitectures. In the proposed method, the search space of the D/VTLM problem can be dramatically reduced through the introduction of the min-max weight maximum-bipartite-matching (MMW-MBM) and a related heuristic bipartite matching method. MMW-MBM is defined on a weighted bipartite graph as an MBM, where the maximal weight of the edges in the matching has a minimal value. In addition, a defect- and variation-aware local search (D/VALS) operator is proposed for D/VTLM and embedded in a global search framework. The D/VALS operator is able to utilize the domain knowledge extracted from problem instances and, thus, has the potential to search the solution space more efficiently. Compared with the state-of-the-art heuristic and recursive algorithms, and a simulated annealing algorithm, the good performance of our proposed method is verified on a 3-bit adder and a large set of random benchmarks of various scales.

The maximum balanced biclique problem (MBBP), an NP-hard combinatorial optimization problem, has been attracting more attention in recent years. Existing node-deletion-based algorithms usually fail to find high-quality solutions due to their easy stagnation in local optima, especially when the scale of the problem grows large. In this paper, a new algorithm for the MBBP, evolutionary algorithm with structure mutation (EA/SM), is proposed. In the EA/SM framework, local search complemented with a repair-assisted restart process is adopted. A new mutation operator, SM, is proposed to enhance the exploration during the local search process. The SM can change the structure of solutions dynamically while keeping their size (fitness) and the feasibility unchanged. It implements a kind of large mutation in the structure space of MBBP to help the algorithm escape from local optima. An MBBP-specific local search operator is designed to improve the quality of solutions efficiently; besides, a new repair-assisted restart process is introduced, in which the Marchiori's heuristic repair is modified to repair every new solution reinitialized by an estimation of distribution algorithm (EDA)-like process. The proposed algorithm is evaluated on a large set of benchmark graphs with various scales and densities. Experimental results show that: 1) EA/SM produces significantly better results than the state-of-the-art heuristic algorithms; 2) it also outperforms a repair-based EDA and a repair-based genetic algorithm on all benchmark graphs; and 3) the advantages of EA/SM are mainly due to the introduction of the new SM operator and the new repair-assisted restart process.

Memetic algorithms (MAs) have been recognized as an effective algorithm framework for solving optimization problems. However, the exiting work mainly focused on the improvement for search operators. Local Search Depth (LSD) is a crucial parameter in MAs, which controls the computing resources assigned for local search. In this paper, an Adaptive Local Search Depth (ALSD) strategy is proposed to arrange the computing resources for local search according to its performance dynamically. A Memetic Algorithm with ALSD (MA-ALSD) is presented, its performance and the effectiveness of ALSD are testified via experiments on the LSGO test suite issued in CEC'2012.

Due to the super scale, high defect density, and per-chip designing paradigm of emerging nanoelectronics, the runtime of the algorithms for defect-tolerant design is of vital importance from the perspective of practicability. In this article, an efficient and effective heuristic defect-free subcrossbar extraction algorithm is proposed which improves performance by mixing the heuristics from two state-of-the-art algorithms and then is speeded up significantly by considerably reducing the number of major loops. Compared with the current most effective algorithm that improves the solution quality (i.e., size of the defect-free subcrossbar obtained) at the cost of high time complexity O ( n 3 ), the time complexity of the proposed heuristic algorithm is proved to be O ( n 2 ). Using a large set of instances of various scales and defect densities, the simulation results show that the proposed algorithm can offer similar high-quality solutions as the current most effective algorithm while consuming much shorter runtimes (reduced to about 1/3 to 1/5) than the current most effective algorithm.

The defect-tolerant logic mapping (DTLM), which has been proved to be an NP-complete combinatorial search problem, is a key step for logic implementation in emerging crossbar-based nano-architectures. However, no practically satisfactory solution has been suggested for the DTLM until now. In this paper, the problem of DTLM is first modeled as a combinatorial optimization problem through the introduction of maximum-bipartite-matching. Then, a new memetic algorithm with fitness approximation (MA/FA) is proposed to solve the optimization problem efficiently. In MA/FA, a new greedy reassignment local search operator, capable of utilizing the domain knowledge and information from problem instances, is designed to help the algorithm find optimal logic mapping with consumption of relatively lower computational resources. A fitness approximation method is adopted to reduce the time consumption of fitness evaluation dramatically. In addition, a hybrid fitness evaluation strategy that combines the exact and approximated fitness evaluation methods is presented to balance the accuracy and time efficiency of fitness evaluation. The effectiveness and efficiency of the proposed methods are testified and evaluated on a large set of benchmark instances of various scales, and the advantage of MA/FA on keeping good balance between effectiveness and efficiency is also observed.

Surrogate models used in evolutionary algorithms (EAs) aim to reduce computationally expensive objective function evaluations. However, low-quality surrogates may mislead EAs and as a result, surrogate-assisted EAs may fail to locate the global optimum. Among various machine learning models for surrogates, Gaussian Process (GP) models have shown to be effective as GP models are able to provide fitness estimation as well as a confidence level. One weakness of GP models is that the computational cost for training increases rapidly as the number of training samples increases. To reduce the computational cost for training, here we propose to adopt an ensemble of local Gaussian Process models. Different from independent local Gaussian Process models, local Gaussian Process models share the same model parameters. Then the performance of the covariance matrix adaptation evolution strategy (CMA-ES) assisted by an ensemble of local Gaussian Process models with five different sampling strategies is compared. Experiments on eight benchmark functions demonstrate that ensembles of local Gaussian Process models can provide reliable fitness prediction and uncertainty estimation. Among the compared strategies, the clustering technique using the lower confidence bound sampling strategy exhibits the best global search performance.

To improve the performance of EAs on large scale numerical optimization problems, a number of techniques have been invented, among which, Cooperative Coevolution (CC in short) is obviously a promising one. But sometimes CC is easy to lead to premature convergence in large scale global optimization. In this paper, a Cooperative Coevolution Evolutionary Algorithm (CCEA in short) with global search (CCGS) is presented to handle large scale global optimization (LSGO) problems. The performance of CCGS is evaluated on the test functions provided for the CEC 2012 competition and special session on Large Scale Global Optimization. The experiment results show that this technique is more effective than CCEAs without global search.

Cooperative Coevolutionary Evolutionary Algorithm is an extension of conventional Evolutionary Algorithm: it implements the idea of divide and conquer by dividing the whole set of variables into several subsets (groups), and evolve each subset independently with a certain optimizer. How to group the variables effectively have been studied by several researchers. Quite a number of variable grouping strategies have been proposed, in most of which, the correlation among variables is considered as the most important factor for guiding grouping, although its legitimacy has not been investigated comprehensively. In this paper an empirical analysis is conducted to testify the legitimacy of assumption that the correlation among variables is an important factor for variable grouping. The experiment results show that, although in some situation, the performance of random grouping is better than that of grouping based on the correct correlation knowledge, the variable correlation is obviously an important factor affecting the performance of the grouping strategies.

Memetic algorithms (MAs) are widely recognized to have better convergence capability than their conventional counterparts. Due to its good robustness and universality, differential evolution (DE) has been frequently used as the global search method in MAs. However, on account of the limited performance of the conventional local search operators, the performance of previous DE-related MAs still needs further improvement. In this paper, we implement more efficient evolutionary algorithms (EAs) as the local search techniques in an adaptive MA framework to form two MA(DE-LS) variants, and investigate their impacts. In order to comprehensively show the effectiveness and efficiency of MA(DE-LS), we experimentally compare it with state-of-the-art EAs, DE-based MAs and other MAs.

During the last decade, a large number of evolutionary algorithmic variants have been proposed for diverse optimization tasks, most of which are practical engineering applications. In the previous research, one variant is always designed for one specific engineering application. In IEEE Congress on Evolutionary Competition, the numerical optimization competition is held to benchmark different optimization algorithms for more general applications. In this paper, we conduct the optimization method estimation of distribution and differential evolution (ED-DE) by implementing a two-stage ensemble idea, whose effectiveness and efficiency has been experimentally verified.

Large-scale global optimization (LSGO) is a very important and challenging task in optimization domain, which is embedded in many scientific and engineering applications. In this paper, a two-stage based ensemble optimization evolutionary algorithm (EOEA) is designed to handle LSGO problems. The performance of EOEA is evaluated on the test functions provided by the LSGO competition of IEEE Congress of Evolutionary Computation (CEC 2010). Compared with some previous LSGO algorithms, EOEA demonstrates better performance.

Economic Load Dispatch (ELD) optimization is an important and difficult task in power system planning. Previously, most of the research mainly focused on proposing various evolutionary algorithms (EAs) to pursue better results of ELD problems. However, few comprehensive analysis of the effects of various constraint handling techniques (CHTs) on the performance of EA-based techniques are available so far. In this paper, we try to fill this gap by experimentally testing the algorithmic variants of combining four effective and widely used EAs with three CHTs. From the experimental results on the ELD problems with valve-point and those problems with both valve-point and multiple-fuel effect, several important conclusions can be achieved, including 1) for the low scale ELD problem with valve-point only, the selection of EAs is more important than CHTs; 2) for the large scale problems, CHTs play crucial roles; 3) the appropriate combination of EA and CHT is helpful to achieve better performance. This study is also expected to provide solid basis for further strengthening the robustness of EAs for ELD optimization. It is also interesting to observe that the experimental results obtained in this paper are much better than those of the previous effective ELD optimization algorithms.

Finger stroke detection is an important topic in hand based Human Computer Interaction (HCI) system. Few research studies have carried out effective solutions to this problem. In this paper, we present a novel approach for stroke detection based on mono vision. Via analyzing the optical flow field within the finger area, our method is able to detect finger stroke under various camera position and visual angles. We present a thorough evaluation for each component of the algorithm, and show its efficiency and effectiveness on solving difficult stroke detection problems.

As the research of dynamic optimization arising, memory-based strategy has gained public attention recently. However, few studies on developing dynamic multi-objective optimization algorithms and even fewer studies on multi-objective memory-based strategy were reported previously. In this paper, we try to address such an issue by proposing several memory-based multi-objective evolutionary algorithms and experimentally investigating different multi-objective dynamic optimization schemes, which include restart, explicit memory, local search memory and hybrid memory schemes. This study is to provide pre-trial research of how to appropriately organize and effectively reuse the changed Pareto-optimal decision values (i.e., Pareto-optimal solutions: POS) information.

Large scale global optimization (LSGO) is a very important and extremely difficult task in optimization domain, which is urgently needed for scientific and engineering applications. Recently, decompose-and-conquer strategy has become a promising method to handle LSGO problems. In this paper, we propose a new strategy variance priority (VP) to improve the classical cooperative co-evolution framework. Based on this proposed strategy, a new LSGO algorithm, variance priority based cooperative co-evolution differential evolution (VP-DECC), is developed. The advantages of VP strategy over the other decompose-and-conquer strategies are experimentally investigated. Especially, it has shown excellent performance in dealing with more complex problems.

Iris localization is a key component of practical iris recognition system. Previous algorithms show good localization performances for iris images captured in the ideal conditions. However, in practice, the quality of iris image is greatly influenced by luminance, eyelashes, hair or glasses frame, which will cause mislocalization. In order to improve the robustness of iris localization, this paper proposes a new localization algorithm based on the radial symmetry transform, in which the radial symmetry characteristic of the pupil is utilized to realize iris localization. Experimental results show that the proposed algorithm can efficiently avoid the interference of luminance and other bad conditions, and realize robust precise localization in a real-time system.

Classical NK-landcape model was designed for analyzing optimization and evolution process in binary solution space, so it can not be used to analyze real coded optimization algorithms (RCOAs) directly, which work in continuous solution space directly. In this paper, the concept of NK-landscape model is extended to the continuous space, and a new NK-landscape model with continuous space is proposed. The new model is powerful and comprehensive with simple structure and flexible formula. Therefore, it can be used to construct test functions of various types of linkages for analyzing various performances of RCOAs. The feasibility of the proposed model is testified via experiments with 3 well-known RCOAs, ( i.e. covariance matrix adapting evolutionary strategy (CMA-ES), differential evolution (DE), neighborhood search differential evolution (NSDE)). The results show that the new model can reveal the merits and demerits of RCOAs effectively.

A univariate EDA denoted as ldquoLSEDA-glrdquo for large scale global optimization (LSGO) problems is proposed in this paper. Three efficient strategies: sampling under mixed Gaussian and Levy probability distribution, standard deviation control strategy and restart strategy are adopted to improve the performance of classical univariate EDA on LSGO problems. The motivation of such work is to extend EDAs to LSGO domain reasonably. Comparison among LSEDA-gl, EDA with standard deviation control strategy only (EDA-STDC) and similar EDA version ldquocontinuous univariate marginal distribution algorithmrdquo UMDAc is carried out on classical test functions. Based on the general comparison standard, the strengths and weaknesses of the algorithms are discussed. Besides, LSEDA-gl is tested on 7 functions with 100, 500, 1000 dimensions provided in the CECpsila2008 Special Session on LSGO. This work is also expected to provide a comparison result for the CECpsila2008 special session.

Hardware–Software (HW–SW) co-synthesis is one of the key steps in modern embedded system design. Generally, HW–SW co-synthesis is to optimally allocate processors, assign tasks to processors, and schedule the processing of tasks to achieve a good balance among performance, cost, power consumption, etc. Hence, it is a typical multi-objective optimization problem. In this paper, a new multi-objective HW–SW co-synthesis algorithm based on the quantum-inspired evolutionary algorithm (MQEAC) is proposed. MQEAC utilizes multiple quantum probability amplitude vectors to model the promising areas of solution space. Meanwhile, this paper presents a new crossover operator to accelerate the convergence to the Pareto front and introduces a PE slot-filling strategy to improve the efficiency of scheduling. Experimental results show that the proposed algorithm can solve the typical multi-objective co-synthesis problems effectively and efficiently.

Hardware-software co-synthesis is a key step of future design of embedded systems. It involves three interdependent subproblems: allocation of resources, assignment of tasks to resources, and scheduling the execution of tasks. Both assignment and scheduling are known to be NP-complete. So it is a really hard and challenging task to optimization algorithms. Both heuristic and evolutionary algorithms are commonly used in real world. Heuristic algorithms converge rapidly but often be trapped in local minima and evolutionary algorithms own high exploration capacity but become time-consuming when handling large-scale systems. In this paper, a new hybrid evolutionary algorithm, called Hybrid Quantum probabilistic coding Genetic Algorithm, is proposed to implement the co-synthesis of large scale multiprocessor embedded systems, in which a heuristic algorithm is combined with the Quantum probabilistic coding Genetic Algorithm to enhance the performance on the hard task. The experimental results show that HQGA has better performance than both HA and QGA on large scale HW/SW co-synthesis problems.