Large Margin Research Articles

Quality Diversity under Sparse Interaction and Sparse Reward: Application to Grasping in Robotics.

Quality-Diversity (QD) methods are algorithms that aim to generate a set of diverse and highperforming solutions to a given problem. Originally developed for evolutionary robotics, most QD studies are conducted on a limited set of domains'mainly applied to locomotion, where the fitness and the behavior signal are dense. Grasping is a crucial task for manipulation in robotics. Despite the efforts of many research communities, this task is yet to be solved. Grasping cumulates unprecedented challenges in QD literature: it suffers from reward sparsity, behavioral sparsity, and behavior space misalignment. The present work studies how QD can address grasping. Experiments have been conducted on 15 different methods on 10 grasping domains, corresponding to 2 different robot-gripper setups and 5 standard objects. The obtained results show that MAP-Elites variants that select successful solutions in priority outperform all the compared methods on the studied metrics by a large margin. We also found experimental evidence that sparse interaction can lead to deceptive novelty. To our knowledge, the ability to efficiently produce examples of grasping trajectories demonstrated in this work has no precedent in the literature.

INTRODUCTION: CLIMATE SUSTAINABILITY – EVIDENCE AND POLICY

Last year saw yet another year of weather extremes. The Copernicus Climate Change Service run by the European Centre for Medium-Range Weather Forecasts on behalf of the European Commission (Copernicus, 2024) measured 2023 as being globally the warmest year since records began in 1850. This was by a large margin (0.17 per cent) over the previous record in 2016, with global surface air temperature at nearly 1.5°C above pre-industrial levels. While last year’s observations embodied an El Niño effect, which every few years sees temperatures affected by warmer waters coming to the surface of the tropical eastern Pacific Ocean, changes and anomalies consistently observed over the last few years across the globe are becoming more pronounced. What is commonly labelled “climate change” is turning into a global climate emergency. No economy or society are immune to its effects. Today, we see the global average temperature at over 1.1°C above pre-industrial levels, a rise that has been extraordinarily rapid on a planetary timescale, and one that has been primarily caused through our (humans) burning fossil fuels. Nearly a decade has passed since the United Nations’ Climate Change Conference in 2015, COP21, where 196 nations adopted The Paris Agreement – a legally binding international treaty on climate change. Its goal was to hold “the increase in the global average temperature to well below 2°C above pre-industrial levels” and to pursue efforts “to limit the increase to 1.5°C”.

A circumpolar review of the breeding distribution and habitat use of the snow petrel (Pagodroma nivea), the world’s most southerly breeding vertebrate

Knowledge of the spatial distribution of many polar seabird species is incomplete due to the remoteness of their breeding locations. Here, we compiled a new database of published and unpublished records of all known snow petrel Pagodroma nivea breeding sites. We quantified local environmental conditions at sites by appending indices of climate and substrate, and regional-scale conditions by appending 30 year mean (1992–2021) sea-ice conditions within accessible foraging areas. Breeding snow petrels are reported at 456 sites across Antarctica and subantarctic islands. Although many counts are old or have large margins of error, population estimates available for 222 known sites totalled a minimum of ~ 77400 breeding pairs. However with so many missing data, the true breeding population will be much higher. Most sites are close to the coast (median = 1.15 km) and research stations (median = 26 km). Median distance to the November sea-ice edge (breeding season sea-ice maximum) is 430 km. Locally, most nests occur in cavities in high-grade metamorphic rocks. Minimum air temperatures occur at inland sites, and maxima at their northern breeding limit. Breeding location and cavity selection is likely determined by availability of suitable breeding substrate within sustainable distance of suitable foraging habitat. Within this range, nest sites may then be selected based on local conditions such as cavity size and aspect. Our database will allow formal analyses of habitat selection and provides a baseline against which to monitor future snow petrel distribution changes in response to climate change.

Encoding matching criteria for cross-domain deformable image registration.

Most existing deep learning-based registration methods are trained on single-type images to address same-domain tasks, resulting in performance degradation when applied to new scenarios. Retraining a model for new scenarios requires extra time and data. Therefore, efficient and accurate solutions for cross-domain deformable registration are indemand. We argue that the tailor-made matching criteria in traditional registration methods is one of the main reason they are applicable in different domains. Motivated by this, we devise a registration-oriented encoder to model the matching criteria of image features and structural features, which is beneficial to boost registration accuracy andadaptability. Specifically, a general feature encoder (Encoder-G) is proposed to capture comprehensive medical image features, while a structural feature encoder (Encoder-S) is designed to encode the structural self-similarity into the global representation. Moreover, by updating Encoder-S using one-shot learning, our method can effectively adapt to different domains. The efficacy of our method is evaluated using MRI images from three different domains, including brain images (training/testing: 870/90 pairs), abdomen images (training/testing: 1406/90 pairs), and cardiac images (training/testing: 64770/870 pairs). The comparison methods include traditional method (SyN) and cutting-edge deep networks. The evaluation metrics contain dice similarity coefficient (DSC) and average symmetric surface distance (ASSD). In the single-domain task, our method attains an average DSC of 68.9%/65.2%/72.8%, and ASSD of 9.75/3.82/1.30 mm on abdomen/cardiac/brain images, outperforming the second-best comparison methods by large margins. In the cross-domain task, without one-shot optimization, our method outperforms other deep networks in five out of six cross-domain scenarios and even surpasses symmetric image normalization method (SyN) in two scenarios. By conducting the one-shot optimization, our method successfully surpasses SyN in all six cross-domainscenarios. Our method yields favorable results in the single-domain task while ensuring improved generalization and adaptation performance in the cross-domain task, showing its feasibility for the challenging cross-domain registration applications. The code is publicly available at https://github.com/JuliusWang-7/EncoderReg.

Research on the Analysis Model of Regional Economic Growth Differences in China Based on Big Data and Intelligent Algorithms

China’s growing economy contributes significantly to worldwide consumption. Economic rebalancing promises opportunities for manufacturing exporters, and that can weaken commodity demand in the long term. China exerts increasing influence on emerging countries through trade, investment, and ideas. China faces several important economic issues that can hinder future development, including distortive economic policies that have evolved into an overreliance on fixed investment and exports for economic growth, government assistance for state-owned businesses, and a weak banking system. Building an information infrastructure and enhancing the technical level and application capability of big data (BD) remain under the purview of the Chinese Government. Artificial intelligence (AI)-based hybrid artificial neural network (HANN) might boost total factor productivity by a large margin, influencing many sectors in China in ways that official statistics would miss, including changes to the labor market, investment patterns, and overall productivity. Hence, BD-HANN has a coefficient of variation, quantity graph analysis, standard deviation, and entropy index, some of the most conventional quantitative research tools used to examine disparities in regional economic growth. According to the neoclassical growth model, regions with lower starting values of the capital-labor ratio anticipate higher per capita income growth rates. Thus, poor areas will grow faster than rich ones, assuming that the only difference between regional economies is the level of their initial stock of capital.

Efficiency and consumer welfare under retail electricity deregulation: Analysis of Ohio's retail choice markets

AbstractMany critical infrastructure services operate under either price regulation or deregulated systems. The grand policy experiment of deregulation has been heavily studied, but evaluations have been limited at the retail level where end‐user prices are often difficult to obtain. This paper presents an in‐depth look at the retail market for electricity in a retail deregulated, or restructured, market—Ohio. We build and introduce a comprehensive SQL database of every daily filed retail electricity offer over a 9‐year period of study, over two‐million records. We integrate this data with other external data sources and conduct a detailed descriptive analysis of market prices as observed by end consumers at the retail level. We find that the lion's share of “competitive” retail electricity choice offers are more costly to consumers than the utility's default service rate and have higher markups over the wholesale price, and we find that when prices exceed the default rate they do so by a considerably larger margin than when consumers observe savings. We also find that even well‐informed consumers are able to find a welfare‐improving rate relative to the default rate between only 43% and 59% of the time. We conclude with a discussion of implications and root economic causes of the efficiency and consumer welfare problems we observe.

Playing for 3D Human Recovery.

Image- and video-based 3D human recovery (i.e., pose and shape estimation) have achieved substantial progress. However, due to the prohibitive cost of motion capture, existing datasets are often limited in scale and diversity. In this work, we obtain massive human sequences by playing the video game with automatically annotated 3D ground truths. Specifically, we contribute GTA-Human, a large-scale 3D human dataset generated with the GTA-V game engine, featuring a highly diverse set of subjects, actions, and scenarios. More importantly, we study the use of game-playing data and obtain five major insights. First, game-playing data is surprisingly effective. A simple frame-based baseline trained on GTA-Human outperforms more sophisticated methods by a large margin. For video-based methods, GTA-Human is even on par with the in-domain training set. Second, we discover that synthetic data provides critical complements to the real data that is typically collected indoor. We highlight that our investigation into domain gap provides explanations for our data mixture strategies that are simple yet useful, which offers new insights to the research community. Third, the scale of the dataset matters. The performance boost is closely related to the additional data available. A systematic study on multiple key factors (such as camera angle and body pose) reveals that the model performance is sensitive to data density. Fourth, the effectiveness of GTA-Human is also attributed to the rich collection of strong supervision labels (SMPL parameters), which are otherwise expensive to acquire in real datasets. Fifth, the benefits of synthetic data extend to larger models such as deeper convolutional neural networks (CNNs) and Transformers, for which a significant impact is also observed. We hope our work could pave the way for scaling up 3D human recovery to the real world.

Towards Visual-Prompt Temporal Answer Grounding in Instructional Video.

Temporal answer grounding in instructional video (TAGV) is a new task naturally derived from temporal sentence grounding in general video (TSGV). Given an untrimmed instructional video and a text question, this task aims at locating the frame span from the video that can semantically answer the question, i.e., visual answer. Existing methods tend to solve the TAGV problem with a visual span-based predictor, taking visual information to predict the start and end frames in the video. However, due to the weak correlations between the semantic features of the textual question and visual answer, current methods using the visual span-based predictor do not work well in the TAGV task. In this paper, we propose a visual-prompt text span localization (VPTSL) method, which introduces the timestamped subtitles for a text span-based predictor. Specifically, the visual prompt is a learnable feature embedding, which brings visual knowledge to the pre-trained language model. Meanwhile, the text span-based predictor learns joint semantic representations from the input text question, video subtitles, and visual prompt feature with the pre-trained language model. Thus, the TAGV is reformulated as the task of the visual-prompt subtitle span localization for the visual answer. Extensive experiments on five instructional video datasets, namely MedVidQA, TutorialVQA, VehicleVQA, CrossTalk and Coin, show that the proposed method outperforms several state-of-the-art (SOTA) methods by a large margin in terms of mIoU score, which demonstrates the effectiveness of the proposed visual prompt and text span-based predictor.

GeoDTR+: Toward Generic Cross-View Geolocalization via Geometric Disentanglement.

Cross-View Geo-Localization (CVGL) estimates the location of a ground image by matching it to a geo-tagged aerial image in a database. Recent works achieve outstanding progress on CVGL benchmarks. However, existing methods still suffer from poor performance in cross-area evaluation, in which the training and testing data are captured from completely distinct areas. We attribute this deficiency to the lack of ability to extract the geometric layout of visual features and models' overfitting to low-level details. Our preliminary work (Zhang et al. 2022) introduced a Geometric Layout Extractor (GLE) to capture the geometric layout from input features. However, the previous GLE does not fully exploit information in the input feature. In this work, we propose GeoDTR+ with an enhanced GLE module that better models the correlations among visual features. To fully explore the LS techniques from our preliminary work, we further propose Contrastive Hard Samples Generation (CHSG) to facilitate model training. Extensive experiments show that GeoDTR+ achieves state-of-the-art (SOTA) results in cross-area evaluation on CVUSA (Workman et al. 2015), CVACT (Liu and Li, 2019), and VIGOR (Zhu et al. 2021) by a large margin (16.44%, 22.71%, and 13.66% without polar transformation) while keeping the same-area performance comparable to existing SOTA. Moreover, we provide detailed analyses of GeoDTR+.

GUESS: GradUally Enriching SyntheSis for Text-Driven Human Motion Generation.

In this article, we propose a novel cascaded diffusion-based generative framework for text-driven human motion synthesis, which exploits a strategy named GradUally Enriching SyntheSis (GUESS as its abbreviation). The strategy sets up generation objectives by grouping body joints of detailed skeletons in close semantic proximity together and then replacing each of such joint group with a single body-part node. Such an operation recursively abstracts a human pose to coarser and coarser skeletons at multiple granularity levels. With gradually increasing the abstraction level, human motion becomes more and more concise and stable, significantly benefiting the cross-modal motion synthesis task. The whole text-driven human motion synthesis problem is then divided into multiple abstraction levels and solved with a multi-stage generation framework with a cascaded latent diffusion model: an initial generator first generates the coarsest human motion guess from a given text description; then, a series of successive generators gradually enrich the motion details based on the textual description and the previous synthesized results. Notably, we further integrate GUESS with the proposed dynamic multi-condition fusion mechanism to dynamically balance the cooperative effects of the given textual condition and synthesized coarse motion prompt in different generation stages. Extensive experiments on large-scale datasets verify that GUESS outperforms existing state-of-the-art methods by large margins in terms of accuracy, realisticness, and diversity.

An Organ-aware Diagnosis Framework for Radiology Report Generation.

Radiology report generation (RRG) is crucial to save the valuable time of radiologists in drafting the report, therefore increasing their work efficiency. Compared to typical methods that directly transfer image captioning technologies to RRG, our approach incorporates organ-wise priors into the report generation. Specifically, in this paper, we propose Organ-aware Diagnosis (OaD) to generate diagnostic reports containing descriptions of each physiological organ. During training, we first develop a task distillation (TD) module to extract organ-level descriptions from reports. We then introduce an organ-aware report generation module that, for one thing, provides a specific description for each organ, and for another, simulates clinical situations to provide short descriptions for normal cases. Furthermore, we design an auto-balance mask loss to ensure balanced training for normal/abnormal descriptions and various organs simultaneously. Being intuitively reasonable and practically simple, our OaD outperforms SOTA alternatives by large margins on commonly used IU-Xray and MIMIC-CXR datasets, as evidenced by a 3.4% BLEU-1 improvement on MIMIC-CXR and 2.0% BLEU-2 improvement on IU-Xray.

Safe Reinforcement Learning With Dual Robustness.

Reinforcement learning (RL) agents are vulnerable to adversarial disturbances, which can deteriorate task performance or break down safety specifications. Existing methods either address safety requirements under the assumption of no adversary (e.g., safe RL) or only focus on robustness against performance adversaries (e.g., robust RL). Learning one policy that is both safe and robust under any adversaries remains a challenging open problem. The difficulty is how to tackle two intertwined aspects in the worst cases: feasibility and optimality. The optimality is only valid inside a feasible region (i.e., robust invariant set), while the identification of maximal feasible region must rely on how to learn the optimal policy. To address this issue, we propose a systematic framework to unify safe RL and robust RL, including the problem formulation, iteration scheme, convergence analysis and practical algorithm design. The unification is built upon constrained two-player zero-sum Markov games, in which the objective for protagonist is twofold. For states inside the maximal robust invariant set, the goal is to pursue rewards under the condition of guaranteed safety; for states outside the maximal robust invariant set, the goal is to reduce the extent of constraint violation. A dual policy iteration scheme is proposed, which simultaneously optimizes a task policy and a safety policy. We prove that the iteration scheme converges to the optimal task policy which maximizes the twofold objective in the worst cases, and the optimal safety policy which stays as far away from the safety boundary. The convergence of safety policy is established by exploiting the monotone contraction property of safety self-consistency operators, and that of task policy depends on the transformation of safety constraints into state-dependent action spaces. By adding two adversarial networks (one is for safety guarantee and the other is for task performance), we propose a practical deep RL algorithm for constrained zero-sum Markov games, called dually robust actor-critic (DRAC). The evaluations with safety-critical benchmarks demonstrate that DRAC achieves high performance and persistent safety under all scenarios (no adversary, safety adversary, performance adversary), outperforming all baselines by a large margin.

SIIR: Symmetrical Information Interaction Modeling for News Recommendation.

Accurate matching between user and candidate news plays a fundamental role in news recommendation. Most existing studies capture fine-grained user interests through effective user modeling. Nevertheless, user interest representations are often extracted from multiple history news items, while candidate news representations are learned from specific news items. The asymmetry of information density causes invalid matching of user interests and candidate news, which severely affects the click-through rate prediction for specific candidate news. To resolve the problems mentioned above, we propose a symmetrical information interaction modeling for news recommendation (SIIR) in this article. We first design a light interactive attention network for user (LIAU) modeling to extract user interests related to the candidate news and reduce interference of noise effectively. LIAU overcomes the shortcomings of complex structure and high training costs of conventional interaction-based models and makes full use of domain-specific interest tendencies of users. We then propose a novel heterogeneous graph neural network (HGNN) to enhance candidate news representation through the potential relations among news. HGNN builds a candidate news enhancement scheme without user interaction to further facilitate accurate matching with user interests, which mitigates the cold-start problem effectively. Experiments on two realistic news datasets, i.e., MIND and Adressa, demonstrate that SIIR outperforms the state-of-the-art (SOTA) single-model methods by a large margin.

One‐Step Van der Waals Integration of Multi‐Threshold 2D Functional Circuits

AbstractTransition‐metal dichalcogenides (TMDs), as atomically thin semiconductors, have shown immense promise for next‐generation electronics due to their high mobility and potential for creating van der Waals heterostructures. However, precise control of the threshold voltage (Vth) in field‐effect transistors (FETs) remains a significant challenge, impeding the development of 2D material circuits with tailored electronic functions. Herein, a graphene‐assisted one‐step van der Waals integration technique is presented for fabricating multi‐threshold 2D functional circuits. Graphene's unique property of having a surface without dangling bonds is utilized as an intermediary layer, allowing for the transfer of electrodes with various work functions and high‐κ dielectric layers onto 2D channel materials in a single step. This technique has successfully produced Al‐gated MoS2 FETs with a Vth of −0.2 V and Au‐gated MoS2 FETs with a Vth of 1.9 V. This precision enables the development of functional devices such as rail‐to‐rail inverters with large noise margins. Furthermore, more complex multi‐threshold 2D circuits are achieved, including basic logic gates (NOT, NAND, NOR), six‐transistor static random‐access memory (6T‐SRAM), and ring oscillators (RO). This work showcases a scalable and effective strategy for threshold voltage engineering in 2D material‐based circuits, paving the way for sophisticated electronic and optoelectronic applications.

주택청약제도의 개선에 관한 연구 - 무순위 청약제도를 중심으로

Unranked subscription refers to a system in which subscriptions and wins are made regardless of the subscription ranking. Depending on its main contents, there are 'unranked post-subscription' methods, 'random supply', and 'resupply contract canceled housing'. These unranked subscriptions do not require a subscription bankbook when accepting subscriptions, and are characterized by the application of exceptions recognized as homeless members, and if there is a large market margin compared to the initial sale price, the subscription competition rate ranges from as few as several thousand to as many as several million. Although one of the purposes of the housing subscription system is to select applicants for sale and supply housing based on the timing of subscription of resident savings products, if you look at the status of unranked subscriptions, it is very difficult to win due to fierce competition, but if you win, it is also called a very good so-called Lotto subscription. In this study, matters related to the improvement of the unranked subscription system were presented. In the case of a large amount of market profit through deliberation, the amount of unranked subscription should be purchased from the public and then sold so that the profit can be used only for residential welfare purposes. Providing the processing of unranked housing to the general public in the form of a sale at a high competition rate would be a more efficient processing method to purchase and process it in the public rather than paying the general subscriber a relative deprivation and a lot of consumptive effects on utility.

Depth-Wise Convolutions in Vision Transformers for efficient training on small datasets

The Vision Transformer (ViT) leverages the Transformer’s encoder to capture global information by dividing images into patches and achieves superior performance across various computer vision tasks. However, the self-attention mechanism of ViT captures the global context from the outset, overlooking the inherent relationships between neighboring pixels in images or videos. Transformers mainly focus on global information while ignoring the fine-grained local details. Consequently, ViT lacks inductive bias during image or video dataset training. In contrast, convolutional neural networks (CNNs), with their reliance on local filters, possess an inherent inductive bias, making them more efficient and quicker to converge than ViT with less data. In this paper, we present a lightweight Depth-Wise Convolution module as a shortcut in ViT models, bypassing entire Transformer blocks to ensure the models capture both local and global information with minimal overhead. Additionally, we introduce two architecture variants, allowing the Depth-Wise Convolution modules to be applied to multiple Transformer blocks for parameter savings, and incorporating independent parallel Depth-Wise Convolution modules with different kernels to enhance the acquisition of local information. The proposed approach significantly boosts the performance of ViT models on image classification, object detection, and instance segmentation by a large margin, especially on small datasets, as evaluated on CIFAR-10, CIFAR-100, Tiny-ImageNet and ImageNet for image classification, and COCO for object detection and instance segmentation. The source code can be accessed at https://github.com/ZTX-100/Efficient_ViT_with_DW.

Electrodeposition of Crystalline p-Si in Molten KF–KCl for Application to Solar Cells

IntroductionElectrodeposition of Si film is a promising manufacturing method to solve the problem of high cost of crystalline silicon solar cells produced by conventional methods. Bard et al. made a crystalline-Si solar cell by a two-step electrodeposition from molten CaCl2–CaO–SiO2 at 1123 K [1]. The p-Si film was firstly electrodeposited onto the graphite plate in the molten salt using Al2O3 or B2O3 as a dopant source. Then, n-Si was electrodeposited onto the obtained p-Si in another molten CaCl2–CaO–SiO2 using Sb2O3 or Ca3(PO4)2 as a dopant source to make a solar cell. The power conversion efficiency (PCE) was reported to be 3.1%. However, the current efficiency of electrodeposition was low and the deposition rate was low, making the process time consuming. Furthermore, they did not report the concentration of dopants, meaning that there is a large margin to improve the PCE by optimizing the concentration of dopants.Our research group has investigated electrodeposition of crystalline Si films in molten KF–KCl using SiCl4 or K2SiF6 as a Si source [2,3]. Smooth Si films have been obtained with high current efficiency and high current density, i.e., high deposition rate. Moreover, we have reported that dense and smooth Si films electrodeposited in KF–KCl–K2SiF6 exhibit n-type semiconductor characteristics, and in KF–KCl–SiCl4 exhibit p-type semiconductor characteristics [4]. Impurities of group 13 and 15 elements are thought to be from the Si source reagents, indicating that the semiconductor properties of the electrodeposited Si films can be controlled if the impurity concentration can be controlled. In order to fabricate Si solar cells with p-n junctions by electrodeposition, it is necessary to find the optimum conditions for electrodeposition of p- and n-Si films. In this study, we investigated the optimal KBF4 concentration to improve the semiconductor properties of p-Si films for solar cells.ExperimentalThe experiments were conducted in KF–KCl–K2SiF6 (KF:KCl = 60:40 mol%, K2SiF6 = 3.5 mol%) at 1023 K in an Ar glovebox. KBF4 as a dopant source was added to the molten salt. The concentration of KBF4 was 0.5, 2, and 10 mol ppm. Graphite plate electrodes were used as the working electrodes. The counter electrode was a glassy carbon rod. The reference electrode was Pt wire. Semiconductor properties of the Si samples prepared by galvanostatic electrolysis were evaluated by linear sweep voltammetry in CH3CN–TBAPF6 (0.1 M)–EV(ClO4)2 (0.05 M) at 293 K under 100 mW cm−2 light with a chopper. The samples were also analyzed by SEM and XRD to observe the morphology and crystallinity of the film.Results and DiscussionFigure 1 shows the optical images and surface SEM images of the samples electrodeposited at 100 mA cm−2 for 180 C cm−2 at 1023 K. The part of graphite plates immersed in molten salt was covered with dark blue electrodeposits. XRD analysis showed that the obtained Si films were well-crystalized Si.Figure 2 shows linear sweep voltammograms of the Si samples and a commercial p-Si wafer in CH3CN–TBAPF6 (0.1 M)–EV(ClO4)2 (0.05 M) at 293 K under light with a chopper. In the case of Si film obtained with 0.5 mol ppm KBF4 addition, no photocurrent was observed, as indicated by the purple line. This is considered to be a compensated semiconductor as the donor and acceptor (added boron) coexist and the effects of the donor and the acceptor cancel each other out. In the photoelectrochemical measurements of Si films obtained with 2 and 10 mol ppm KBF4 addition and a commercial p-Si wafer, stepwise currents were observed corresponding to the chopping of light, as shown by the red, blue, and black lines. This indicates the presence of photoexcited electrons, confirming that it is p-Si. However, the Si film obtained with 10 mol ppm KBF4 addition had a larger dark current (current when no light is irradiated), indicating that the concentration of KBF4 was too high. Taking these into consideration, it is thought that the optimal concentration of KBF4 addition is around 2 mol ppm.ReferencesZou, L. Ji, J. Ge, D. R. Sadoway, E. T. Yuand A. J. Bard, Nat Commun, 10, 5772 (2019).Maeda, K. Yasuda, T. Nohira, R. Hagiwara, and T. Homma, J. Electrochem. Soc.,162, D444 (2015).Yasuda, K. Maeda, R. Hagiwara, T. Homma, and T. Nohira, J. Electrochem. Soc., 164, D67 (2017).Yasuda, T. Kato, Y. Norikawa, and T. Nohira, J. Electrochem. Soc., 168, 112502 (2021). Figure 1

Multi-Modal CLIP-Informed Protein Editing.

Background: Proteins govern most biological functions essential for life, and achieving controllable protein editing has made great advances in probing natural systems, creating therapeutic conjugates, and generating novel protein constructs. Recently, machine learning-assisted protein editing (MLPE) has shown promise in accelerating optimization cycles and reducing experimental workloads. However, current methods struggle with the vast combinatorial space of potential protein edits and cannot explicitly conduct protein editing using biotext instructions, limiting their interactivity with human feedback. Methods: To fill these gaps, we propose a novel method called ProtET for efficient CLIP-informed protein editing through multi-modality learning. Our approach comprises 2 stages: In the pretraining stage, contrastive learning aligns protein-biotext representations encoded by 2 large language models (LLMs). Subsequently, during the protein editing stage, the fused features from editing instruction texts and original protein sequences serve as the final editing condition for generating target protein sequences. Results: Comprehensive experiments demonstrated the superiority of ProtET in editing proteins to enhance human-expected functionality across multiple attribute domains, including enzyme catalytic activity, protein stability, and antibody-specific binding ability. ProtET improves the state-of-the-art results by a large margin, leading to substantial stability improvements of 16.67% and 16.90%. Conclusions: This capability positions ProtET to advance real-world artificial protein editing, potentially addressing unmet academic, industrial, and clinical needs.

GANzzle[formula omitted]: Generative approaches for jigsaw puzzle solving as local to global assignment in latent spatial representations

Jigsaw puzzles are a popular and enjoyable pastime that humans can easily solve, even with many pieces. However, solving a jigsaw is a combinatorial problem, and the space of possible solutions is exponential in the number of pieces, intractable for pairwise solutions. In contrast to the classical pairwise local matching of pieces based on edge heuristics, we estimate an approximate solution image, i.e., a mental image, of the puzzle and exploit it to guide the placement of pieces as a piece-to-global assignment problem. Therefore, from unordered pieces, we consider conditioned generation approaches, including Generative Adversarial Networks (GAN) models, Slot Attention (SA) and Vision Transformers (ViT), to recover the solution image. Given the generated solution representation, we cast the jigsaw solving as a 1-to-1 assignment matching problem using Hungarian attention, which places pieces in corresponding positions in the global solution estimate. Results show that the newly proposed GANzzle-SA and GANzzle-VIT benefit from the early fusion strategy where pieces are jointly compressed and gathered for global structure recovery. A single deep learning model generalizes to puzzles of different sizes and improves the performances by a large margin. Evaluated on PuzzleCelebA and PuzzleWikiArts, our approaches bridge the gap of deep learning strategies with respect to optimization-based classic puzzle solvers.

On-line image analysis for evaporative crystallization of xylose

On-line imaging technique was investigated for monitoring industrial production of xylose crystals via evaporation, with the focus being on the fine crystals since they affect the efficacy of downstream filtration, and also lead to product loss. High solid concentration and crystal agglomeration pose great challenges to the image analysis, hindering quantitative understanding of the crystallization process and possible control actions. Three different analysis methods were examined, including deep learning-based semantic segmentation (Swin Transformer) and instance segmentation (Mask R-CNN) together with an advanced orthodox approach. Swin Transformer outperformed other two methods by a large margin, indicating that the semantic segmentation is more suitable for the present scenario of evaporation crystallization. Its precision, recall, F1 score, Jaccard index, and accuracy reach 0.903, 0.887, 0.894, 0.810, and 0.873, respectively. The fine crystals were extracted by image post-processing, and the evolution of their area fractions, equivalent diameters, and aspect ratios were analyzed in details.