Recently, deep unfolding networks (DUNs) have emerged as a promising technique for image Compressive Sensing (CS) reconstruction by unfolding optimization algorithms, where each stage of the DUNs corresponds to an iteration of the optimization algorithm. DUNs can be divided into convex optimization based methods and non-convex optimization based methods. On the one hand, DUNs based on convex optimization algorithms cannot handle non-convex optimization problems, thereby limiting their use when the prior term is a non-convex function. On the other hand, although DUNs based on non-convex optimization algorithms can handle more complex prior terms to make global optimal solutions closer to the ground truth, there is a high probability that they converge only to a local optimum. Therefore, in practical applications, it is necessary to consider the various characteristics of the problem comprehensively, then design appropriate prior terms and choose convex or non-convex optimization in DUN. This paper proposes ViP-DUN method to learn suitable prior terms and adaptively use convex or non-convex optimization. ViP-DUN learns deep prior terms and variable metrics in a data-driven manner to achieve adaptive use of convex or non-convex optimization. Moreover, we designed a lightweight multi-scale information fusion module in ViP-DUN at the network structure level to further enhance the network's processing capability. Experiments demonstrate that our proposed method can improve image reconstruction quality at multiple compression rates through the adaptive capabilities of the network. Our complete code will be made publicly available upon acceptance.

Recently, Deep Unfolding Networks (DUNs) have achieved impressive reconstruction quality in the field of image Compressive Sensing (CS) by unfolding iterative optimization algorithms into neural networks. The reconstruction quality of DUNs depends on the learned prior knowledge, so introducing stronger prior knowledge can further improve reconstruction quality. On the other hand, pre-trained diffusion models contain powerful prior knowledge and have a solid theoretical foundation and strong scalability, but it requires a large number of iterative steps to achieve reconstruction. In this paper, we propose to use the powerful prior knowledge of pre-trained diffusion model in DUNs to achieve high-quality reconstruction with less steps for image CS. Specifically, we first design an iterative optimization algorithm named Diffusion Message Passing (DMP), which embeds a pre-trained diffusion model into each iteration process of DMP. Then, we deeply unfold the DMP algorithm into a neural network named DMP-DUN. The proposed DMP-DUN can use lightweight neural networks to achieve mapping from measurement data to the intermediate steps of the reverse diffusion process and directly approximate the divergence of the diffusion model, thereby further improving reconstruction efficiency. Extensive experiments show that our proposed DMP-DUN achieves state-of-the-art performance and requires at least only 2 steps to reconstruct the image. Codes are available at https://github.com/FengodChen/DMP-DUN-CVPR2025.

Social interaction between microbes can be described at many levels of details: from the biochemistry of cell-cell interactions to the ecological dynamics of populations. Choosing an appropriate level to model microbial communities without losing generality remains a challenge. Here we show that modeling cross-feeding interactions at an intermediate level between genome-scale metabolic models of individual species and consumer-resource models of ecosystems is suitable to experimental data. We applied our modeling framework to three published examples of multi-strain Escherichia coli communities with increasing complexity: uni-, bi-, and multi-directional cross-feeding of either substitutable metabolic byproducts or essential nutrients. The intermediate-scale model accurately fit empirical data and quantified metabolic exchange rates that are hard to measure experimentally, even for a complex community of 14 amino acid auxotrophies. By studying the conditions of species coexistence, the ecological outcomes of cross-feeding interactions, and each community's robustness to perturbations, we extracted new quantitative insights from these three published experimental datasets. Our analysis provides a foundation to quantify cross-feeding interactions from experimental data, and highlights the importance of metabolic exchanges in the dynamics and stability of microbial communities.

Pseudomonas aeruginosa, a main cause of human infection, can gain resistance to the antibiotic aztreonam through a mutation in NalD, a transcriptional repressor of cellular efflux. Here we combine computational analysis of clinical isolates, transcriptomics, metabolic modeling and experimental validation to find a strong association between NalD mutations and resistance to aztreonam-as well as resistance to other antibiotics-across P. aeruginosa isolated from different patients. A detailed analysis of one patient's timeline shows how this mutation can emerge in vivo and drive rapid evolution of resistance while the patient received cancer treatment, a bone marrow transplantation, and antibiotics up to the point of causing the patient's death. Transcriptomics analysis confirmed the primary mechanism of NalD action-a loss-of-function mutation that caused constitutive overexpression of the MexAB-OprM efflux system-which lead to aztreonam resistance but, surprisingly, had no fitness cost in the absence of the antibiotic. We constrained a genome-scale metabolic model using the transcriptomics data to investigate changes beyond the primary mechanism of resistance, including adaptations in major metabolic pathways and membrane transport concurrent with aztreonam resistance, which may explain the lack of a fitness cost. We propose that metabolic adaptations may allow resistance mutations to endure in the absence of antibiotics and could be targeted by future therapies against antibiotic resistant pathogens.

This study suggests a new keyboard layout to efficiently type in Chinese characters. The layout was based on a statistical analysis of Chinese corpus and was derived from a genetic algorithm. In the semantic analysis, 6 million Chinese characters were transcribed into typing script. Macros of polyphone were disambiguated from either the context or the presence probabilities in the training data. Relative frequencies of single letter and letter pair were counted to investigate on tapping workload and sequence. In the genetic algorithm, five ergonomics criteria—tapping workload distribution, hand alternation, finger alternation, avoidance of big steps, and hit direction—were applied to evaluate keyboard layout alternatives. The result showed that the proposed layout is 43% better than the QWERTY layout in terms of the weighted sum of the five ergonomic criteria.

Customisation of mobile phones is a process of producing products according to individual needs on design, cost, and easiness of the phones. With the aim of identifying the most important features in customising mobile phones, 288 questionnaires were collected and analysed. The result showed that ‘text message’, ‘battery’, ‘contacts’, ‘software updates’, and ‘display size’ were highly required to customise. Among six factors (physical design, technical design, cost of entertainment, cost of information, cost of durability, easiness of use) obtained from a factor analysis, the most important reason for users to customise mobile phones was that they wanted to use a mobile phone easily. Cost of durability and cost of information were also important motivations for customisation of mobile phones. Finally, this research showed that gender and user experience were significant factors for customisation.

As VLSI technology enters the nanoscale regime, design reliability is becoming increasingly important. A major design reliability concern arises from electromigration which refers to the transport of material caused by ion movement in interconnects. Since the lifetime of an interconnect drastically depends on the current flowing through it, the electromigration problem aggravates with increasingly growing thinner wires. Further, the current-density-induced interconnect thermal issue becomes much more severe with larger current. To mitigate the electromigration and the current-density-induced thermal effects, interconnect current density needs to be reduced. Assigning wires to thick metals increases wire volume, and thus, reduces the current density. However, overstretching thick-metal assignment may hurt routability. Thus, it is highly desirable to minimize the thick-metal usage, or total wire cost, subject to the reliability constraint. In this paper, the minimum cost reliability-driven routing, which consists of Steiner tree construction and layer assignment, is considered. The problem is proven to be NP-hard and a highly effective iterative rounding-based integer linear programming algorithm is proposed. In addition, a unified routing technique is proposed to directly handle multiple current levels, which is critical in analog VLSI design. Further, the new algorithm is extended to handle blockage. Our experiments on 450 nets demonstrate that the new algorithm significantly outperforms the state-of-the-art work [CHECK END OF SENTENCE] with up to 14.7 percent wire reduction. In addition, the new algorithm can save 11.4 percent wires over a heuristic algorithm for handling multiple currents.

With advancing technology, large dynamic power consumption has significantly limited circuit miniaturization. Minimizing peak power consumption, which is defined as the maximum power consumption among all voltage partitions, is important since it enables energy saving from the voltage island shutdown mechanism. In this paper, we prove that the peak power driven voltage partitioning problem is NP-complete and propose an efficient provably good fully polynomial time approximation scheme for it. The new algorithm can approximate the optimal peak power driven voltage partitioning solution in O(m2(mn/ϵ4)) time within a factor of (1 + ϵ) for sufficiently small positive e, where n is the number of circuit blocks and m is the number of partitions which is a small constant in practice. Our experimental results demonstrate that the dynamic programming cannot finish for even 20 blocks while our new approximation algorithm runs fast. In particular, varying e, orders of magnitude speedup can be obtained with only 0.6% power increase. The tradeoff between the peak power minimization and the total power minimization is also investigated. We demonstrate that the total power minimization algorithm obtains good results in total power but with quite large peak power, while our peak power optimization algorithm can achieve on average 26.5% reduction in peak power with only 0.46% increase in total power. Moreover, our peak power driven voltage partitioning algorithm is integrated into a simulated annealing based floorplanning technique. Experimental results demonstrate that compared to total power driven floorplanning, the peak power driven floorplanning can significantly reduce peak power with only little impact in total power, HPWL, estimated power ground routing cost, level shifter cost and runtime. Further, when the voltage island shutdown is performed, peak power driven voltage partitioning can lead to over 10% more energy saving than a greedy frequency based voltage partitioning when multiple idle block sequences are considered.

Microarray data typically have thousands of genes, and thus feature extraction is a critical problem for accurate cancer classification. In this paper, a feature extraction method based on the discrete wavelet transform (DWT) is proposed. The approximation coefficients of DWT, together with some useful features from the high-frequency coefficients selected by the maximum modulus method, are used as features. The combined coefficients are then forwarded to a SVM classifier. Experiments are performed on two standard benchmark data sets: ALL/AML Leukemia and Colon tumor. Experimental results show that the proposed method can achieve state-of-the-art performance on cancer classification.