General Process Research Articles

RealtimeGen: An Intervenable AI Image Generation System for Commercial Digital Art Asset Creators

Recent advances in artificial intelligence-generated content (AIGC) have led to the rapid generation of high-quality images. AIGC has attracted the attention of commercial digital art asset creators. Traditional artist-led processes contrast with current AI tools that often reduce creators to passive roles. This study examines the integration of AI image generation into commercial digital art, emphasizing the importance of preserving creators’ creative autonomy. Our formative study (S1) involved interviews with commercial digital art creators, highlighting a need for greater control and transparency in AI-assisted painting. In response, we developed RealtimeGen, an integrated tool that merges human creativity with AI’s capabilities, allowing creators to intervene in the generative process. A user study (S2) comparing RealtimeGen with the popular AIGC tool Stable Diffusion was also carried out. The results showed its enhanced user experience and workflow compatibility. Our work contributes to understanding and improving AI-assisted painting workflows for commercial creators, offering them greater creative agency.

Enhancing Coherence and Diversity in Multi-class Slogan Generation Systems

Many problems related to natural language processing are solved by neural networks and big data. Researchers have previously focused on single-task supervised goals with limited data management to train slogan classification. A multi-task learning framework is used to learn jointly across several tasks related to generating multi-class slogan types. This study proposes a multi-task model named slogan generative adversarial network systems (Slo-GAN) to enhance coherence and diversity in slogan generation, utilizing generative adversarial networks and recurrent neural networks (RNN). Slo-GAN generates a new text slogan-type corpus, and the training generalization process is improved. We explored active learning (AL) and meta-learning (ML) for dataset labeling efficiency. AL reduced annotations by 10% compared to ML but still needed about 70% of the full dataset for baseline performance. The whole framework of Slo-GAN is supervised and trained together on all of these tasks. The text with the higher reporting score level is filtered by Slo-GAN, and a classification accuracy of 87.2% is achieved. We leveraged relevant datasets to perform a cross-domain experiment, reinforcing our assertions regarding both the distinctiveness of our dataset and the challenges of adapting bilingual dialects to one another.

Creating and Reliving the Moment: Using Musical Improvisation and Care Aesthetics as a Lens of Connection and Self-Expression for Younger People Living with Dementia.

Musical improvisation is a generative process of spontaneously creating music 'in the moment'. For people with young onset dementia, musical improvisation provides an extended opportunity for creative self-expression and connection to one's own body and life story. Using visual research methods, including video elicitation interviews, this paper explores the 'in the moment' musical experiences of five people living with young onset dementia who took part in a 15-week improvised music-making programme (Music in Mind). We frame the exploration of the group's musical experiences through the emerging lens of 'care aesthetics'-a concept that identifies the sensory relations and embodied practices between two (or more) people in a caring relationship. In the context of this analysis, we look to the caring practices by, with, and between people living with dementia, their family members, and the musicians who lead the programme and the relationship of these practices to feelings of self-expression and meaningful connection. Musical improvisation has the potential to support the psychological, social, and spiritual wellbeing of people living with young onset dementia. In applying a lens of care aesthetics, it is possible to observe the micro-level experiences of people living with dementia and their family carers.

Intragroup Processes in Social Psychology: My Past, Our Present, Your Future

The first section of this article summarizes my life as a group researcher, beginning with my graduate student years at the University of Wisconsin and continuing with my time as a faculty member at the University of Pittsburgh, where my work focused on majority/minority disagreement, with an emphasis on reaction to deviance group socialization and the relationship between social and cognitive processes. The second section presents my analysis of the decline of intragroup process research in social psychology and reasons why such research is so much more popular in organizational psychology/behavior. The third section lays out a possible path to a brighter future in which social psychologists focus on how generic intragroup processes operate in important real-world groups and employ multiple methodologies to increase the external validity of their research. In this context, I suggest that families and extreme groups are particularly worthy of attention.

Exploration of regularities in bipartite graphs using GEOGEBRA software

A classroom proposal is presented to integrate contents of Graph Theory and Linear Algebra in complete bipartite graphs, linking adjacency and Laplacian matrices, the eigenvalues of graphs will be determined, applicable to connectivity concepts. Students will be given exploration activities working with GeoGebra software, starting from several particular cases, with table works and questionnaires to be completed, in order to determine patterns on the eigenvalues of adjacency and Laplacian matrices of complete bipartite graphs. The work with patterns will lead to the generalization process, to abstract properties from observation and experimentation on examples. This learning experience builds bridges between the concrete and the symbolic, and the student is initiated in research

Denoising diffusion post-processing for low-light image enhancement

Low-light image enhancement (LLIE) techniques attempt to increase the visibility of images captured in low-light scenarios. However, as a result of enhancement, a variety of image degradations such as noise and color bias are revealed. Furthermore, each particular LLIE approach may introduce a different form of flaw within its enhanced results. To combat these image degradations, post-processing denoisers have widely been used, which often yield oversmoothed results lacking detail. We propose using a diffusion model as a post-processing approach, and we introduce Low-light Post-processing Diffusion Model (LPDM) in order to model the conditional distribution between under-exposed and normally-exposed images. We apply LPDM in a manner which avoids the computationally expensive generative reverse process of typical diffusion models, and post-process images in one pass through LPDM. Extensive experiments demonstrate that our approach outperforms competing post-processing denoisers by increasing the perceptual quality of enhanced low-light images on a variety of challenging low-light datasets. Source code is available at https://github.com/savvaki/LPDM.

Stabilized GAN models training with kernel-histogram transformation and probability mass function distance

Image generation using generative adversarial networks (GANs) has been extensively researched in recent years. Despite active developments, the chronic issue of training instability in GANs remains unresolved. To alleviate this problem, this study proposes a model named probability mass function GANs (PMF-GAN), which handles the inherent limitation of GANs. The PMF-GAN framework employs kernels, histogram transformation, and probability mass function (PMF) distance for distribution learning. The configuration of PMF-GAN kernel and PMF distance offers flexibility, allowing for optimal settings tailored to datasets and experimental environments. In this study, experiments were conducted using the gaussian kernel across five different distances. The experiments demonstrated that PMF-GAN outperforms the baselines in terms of visual quality and evaluation metrics, such as Inception score and Frechet Inception distance (FID). For example, in the CIFAR-10 dataset, Euclidean-based PMF-GAN applying with 3 bins showed a 21.5 % and 32.8 % improvement in Inception score and FID, respectively, compared to conventional WGAN-GP. Similarly, in the AFHQ dataset with the same settings, the improvements were 56.9 % and 61.5 %. As a result, this study presents the potential to achieve stable training processes in GAN models with modified loss function structures. The flexibility of the proposed model allows for simultaneous application to various models, contributing to the overall improvement of generative model training processes in the future.

Structure-aware diffusion for low-dose CT imaging

Reducing the radiation dose leads to the x-ray computed tomography (CT) images suffering from heavy noise and artifacts, which inevitably interferes with the subsequent clinic diagnostic and analysis. Leading works have explored diffusion models for low-dose CT imaging to avoid the structure degeneration and blurring effects of previous deep denoising models. However, most of them always begin their generative processes with Gaussian noise, which has little or no structure priors of the clean data distribution, thereby leading to long-time inference and unpleasant reconstruction quality. To alleviate these problems, this paper presents a Structure-Aware Diffusion model (SAD), an end-to-end self-guided learning framework for high-fidelity CT image reconstruction. First, SAD builds a nonlinear diffusion bridge between clean and degraded data distributions, which could directly learn the implicit physical degradation prior from observed measurements. Second, SAD integrates the prompt learning mechanism and implicit neural representation into the diffusion process, where rich and diverse structure representations extracted by degraded inputs are exploited as prompts, which provides global and local structure priors, to guide CT image reconstruction. Finally, we devise an efficient self-guided diffusion architecture using an iterative updated strategy, which further refines structural prompts during each generative step to drive finer image reconstruction. Extensive experiments on AAPM-Mayo and LoDoPaB-CT datasets demonstrate that our SAD could achieve superior performance in terms of noise removal, structure preservation, and blind-dose generalization, with few generative steps, even one step only.

HYPNOINDUCTORS OF THE EFFECTIVE FORMULA OF POSITIVATION IN SOCIAL ADVERTISING DISCOURSE

The article is devoted to influential elements of social advertising discourse, where Milton-model inductions are arranged, which is proposed to be nominated as hypnoinducers. The purpose of the study was to identify actual hypnoinducers of the influential formula of positivity in social advertising discourse. The object of intelligence research is Ukrainian advertising megadiscourse, the subject is suggestive markers of Ukrainian social advertising discourse. General scientific methods contributed to the delineation of specific features of social advertising discourse, various types of hypnoinducers (descriptive method); identification of the components of a communicative suggestion and combining them into a whole (methods of analysis and synthesis); concretization of general conclusions (method of induction); construction of formula models of advertising influence (modeling method). Special linguistic — the study of the semantic load and functional features of linguistic components of social advertising discourse (contextualinterpretive and component analysis) and the communicative nature of advertising suggestogens (elements of discursive analysis). The methodological base of the latest scientific directions (suggestive linguistics and neurolinguistic programming) made it possible to differentiate and systematize relevant markers (hypnoinducers) of advertising suggestion (methods of meta- and Milton-model identification and suggestive qualification, which are distinguished by R. Bandler, J. Grinder, M. Erikson, T. Yu. Kovalevska. The formulas of vectors of emotional orientation of influence are presented: positivity, negation and change of vector (in commercial, political and social advertising discourses), where special attention is paid to the formula of positivity realized in social advertising discourse. Relevant hypnoinducers, which are fixed in the above formula. In particular, hypnoinducers of complex syntactic structure + lexical-semantic hypnoinductors ("point" markers) and actually lexical-semantic hypnoinducers are singled out. Truisms can act as hypnoinducers of complex syntactic structure; pattern break; "reading thoughts" complicated by other hypnoinducers, and quotations, or "quotes" of an established and transformed structure, which, in turn, often act as separate hypnoinducers without the support of lexical and semantic markers. Elements of the inverse metamodel of the processes of omission (nominalization, nonspecific vocabulary) and generalization (universal quantifiers, modal operators of necessity) act as lexical-semantic hypnoinducers. The characteristic feature is the presence of affectives as super (powerful) hypnoinducers, the presence of ergonomics varies, and predicates are not clearly represented. Prospects for research development: in-depth analysis and systematization of current hypnoinducers in various genres of advertising discourse, which will expand the spectrum of the most effective influential dominants.

An integrative approach to protein sequence design through multiobjective optimization.

With recent methodological advances in the field of computational protein design, in particular those based on deep learning, there is an increasing need for frameworks that allow for coherent, direct integration of different models and objective functions into the generative design process. Here we demonstrate how evolutionary multiobjective optimization techniques can be adapted to provide such an approach. With the established Non-dominated Sorting Genetic Algorithm II (NSGA-II) as the optimization framework, we use AlphaFold2 and ProteinMPNN confidence metrics to define the objective space, and a mutation operator composed of ESM-1v and ProteinMPNN to rank and then redesign the least favorable positions. Using the two-state design problem of the foldswitching protein RfaH as an in-depth case study, and PapD and calmodulin as examples of higher-dimensional design problems, we show that the evolutionary multiobjective optimization approach leads to significant reduction in the bias and variance in RfaH native sequence recovery, compared to a direct application of ProteinMPNN. We suggest that this improvement is due to three factors: (i) the use of an informative mutation operator that accelerates the sequence space exploration, (ii) the parallel, iterative design process inherent to the genetic algorithm that improves upon the ProteinMPNN autoregressive sequence decoding scheme, and (iii) the explicit approximation of the Pareto front that leads to optimal design candidates representing diverse tradeoff conditions. We anticipate this approach to be readily adaptable to different models and broadly relevant for protein design tasks with complex specifications.

A prospective multi-center study quantifying visual inattention in delirium using generative models of the visual processing stream

The visual attentional deficits in delirium are poorly characterized. Studies have highlighted neuro-anatomical abnormalities in the visual processing stream but fail at quantifying these abnormalities at a functional level. To identify these deficits, we undertook a multi-center eye-tracking study where we recorded 210 sessions from 42 patients using a novel eye-tracking system that was made specifically for free-viewing in the (ICU); each session lasted 10 min and was labeled with the delirium status of the patient using the Confusion Assessment Method in ICU (CAM-ICU). To analyze this data, we formulate the task of visual attention as a hierarchical generative process that yields a probabilistic distribution of the location of the next fixation. This distribution can then be compared to the measured patient fixation producing a correctness score which is tallied compared across delirium status. This analysis demonstrated that the visual processing system of patients suffering from delirium is functionally restricted to a statistically significant degree. This is the first study to explore the potential mechanisms underpinning visual inattention in delirium and suggests a new target of future research into a disease process that affects one in four hospitalized patients with severe short and long-term consequences.

PMSGAN: Parallel Multistage GANs for Face Image Translation.

In this article, we address the face image translation task, which aims to translate a face image of a source domain to a target domain. Although significant progress has been made by recent studies, face image translation is still a challenging task because it has more strict requirements for texture details: even a few artifacts will greatly affect the impression of generated face images. Targeting to synthesize high-quality face images with admirable visual appearance, we revisit the coarse-to-fine strategy and propose a novel p arallel m ultistage architecture on the basis of g enerative a dversarial n etworks (PMSGAN). More specifically, PMSGAN progressively learns the translation function by disintegrating the general synthesis process into multiple parallel stages that take images with gradually decreasing spatial resolution as inputs. To prompt the information exchange between various stages, a cross-stage atrous spatial pyramid (CSASP) structure is specially designed to receive and fuse the contextual information from other stages. At the end of the parallel model, we introduce a novel attention-based module that leverages multistage decoded outputs as in situ supervised attention to refine the final activations and yield the target image. Extensive experiments on several face image translation benchmarks show that PMSGAN performs considerably better than state-of-the-art approaches.

Human’s life-world and their worldview: ontological-psychological mechanisms of interaction

The ontological characteristics of the worldview are outlined based on the propositions about the connection between man and the world, their systemic interaction, the determining role of life relationships in the formation of the life world. The features of human interaction with the world as two open systems, which simultaneously act as independent systems and as subsystems of each other, are considered: determinism/self-determination, openness/closedness, involvement/distance. It is emphasized that the interaction within the “Man-World” system takes place in the form of life relations, which are manifested in four main modes (as a relation to the objective world, to another person, to the Absolute, to ones own self) and corresponding meeans of subjects interaction with an object (activity, dialogic relations, self-determination in relation to existential meanings and higher values, cognition and expression of ones own self). Establishing life relations in a certain way and in a certain mode, person thereby builds his/her life world a specially organized reality formed by the dynamic systemic unity of the inner world of a person and that part of the objectively existing external world with which he/she is connected by life relations. Additional possibilities for the analysis of the life world are provided by considering it as an ideal model, which carries knowledge about the utilitarian-pragmatic and symbolic characteristics of social and natural objects, as well as about the corresponding mental states of the subject-carrier of this model. The specific features of such knowledge are the intuitively obvious forms of its provision in everyday human experience, the ability to represent universal cultural meanings and effectiveness. In the plane of objective human relations with the world worldview turns out to be a special way of including a person into the world: on the one hand, it mediates (by posing and searching for answers to “big questions”) the processes of constructing a persons own life world, determining its ontological, epistemological, axiological and praxeological boundaries; on the other hand, it creates contexts in which the external world is transformed (through dialogue-based processes of evaluation, idealization, and generalization) into images, meanings, and values (ideals, beliefs, and life principles) that are existentially significant for a person.

Adding a scholarly analysis of teaching and learning to SoTL: the development of the hands-on Utrecht Roadmap for SoTL

ABSTRACT This article outlines the development of a tool designed to enhance Scholarship of Teaching and Learning (SoTL) endeavours: the Utrecht Roadmap for Scholarship of Teaching and Learning (UR-SoTL). Integral to this tool is the incorporation of the Context-Intervention-Mechanism-Outcome (CIMO) logic model, offering a nuanced analysis of the teaching and learning context within the conventional SoTL steps. The CIMO-logic model aims to guide both novice and experienced SoTL practitioners in scrutinizing their teaching activities within the framework of generative learning processes described in relevant literature. This approach enables the scholarly analysis that directs SoTL-projects towards formulating well-grounded and focused research questions and plan-of-approaches.

브랜드 로고 디자인을 위한 생성형 AI 파인튜닝 프로세스 설계 - 달리(DALL-E) 활용을 중심으로

브랜드 로고 디자인을 위한 생성형 AI 파인튜닝 프로세스 설계 - 달리(DALL-E) 활용을 중심으로

Evaluating the Geometry of Objects in Cartographic Generalization with Hu's Invariants

ABSTRACT This paper presents a novel approach to assessing the geometry of objects using Hu's invariants in the context of cartographic generalization. The primary focus is to improve the generalization process and produce more readable and informative maps. The study demonstrates the applicability and effectiveness of the modified invariant moment M1* in evaluating regular shape similarity. Experiments, based on 24 shapes, exhibit greater stability in the results and reveal the high suitability of this moment in the investigation and classification of buildings, among other generalization processes. The efficiency of the proposed method is compared to previous generalization techniques, showing a significant improvement in the generalization process. In conclusion, this research contributes to the development of cartographic generalization methods by introducing the use of Hu's invariants for evaluating object geometry. This approach can improve the automation of map generalization processes and more effective communication of geographic information.

On the shoreline monitoring via earth observation: An isoradiometric method

Shoreline variations, triggered by climate change, eustatism, and tectonic, drive the coastal landscape evolution over multiple spatial and temporal scales. Among the many different existing coast types, sandy coasts are the most sensitive to coastal erosion and accretion processes and, at the same time, often host valuable anthropogenic assets. The rapid and ongoing evolution of these coastal environments poses challenges for their management, necessitating cost-effective and highly reliable methods for measuring these changes. Many remotely sensed shoreline extraction methods have been proposed in the literature, providing valuable tools for improving coastal management. Even if these methodologies allow the demarcation of the shoreline, its pixelated shape usually requires refinement through subsequent smoothing or vector generalization processes. It is important to note that the position of the thus extracted coastline is not a direct result of a measured physical quantity but rather a product of these refinement techniques. To address this problem, we developed a sub-pixel resolution method for extracting shorelines from remotely sensed images of sandy beaches, leveraging the radiometric signature of the shoreline. Validated through precise Global Navigation Satellite System field surveys for positioning the beach foreshore, this method was successfully applied to three beaches in Sicily, in the central Mediterranean, all exhibiting similar microtidal conditions. Its robust design allows for application across various satellite images, employing a straightforward radiometric interpolation method adaptable to different spatial resolutions. This method would be a valuable tool for coastal managers in detecting and mitigating coastal erosion and developing and maintaining anthropogenic coastal assets.

A linguistic approach to Natural Language Generation: The language spectrum, language models, and OpenLogos

Natural Language Generation, a well-established research area within the field of Natural Language Processing, has achieved significant milestones such as in machine translation and paraphrasing. Large Language Models, while improving text quality, show varying effectiveness across different linguistic registers and cultural contexts, raising concerns mostly due to (i) difficulties in understanding the internal workings of the models; (ii) difficulties in understanding the generative processes, which are not transparent; (iii) difficulties in human intervention in the generative function; (iv) propensity for hallucination and incorrect information content, (v) potential irresponsible use of extensive resources from unidentified sources, and (vi) the risk of misuse, among the most obvious hurdles. This article has three main goals: (1) suggest a linguistic approach to Natural Language Generation, (2) explore the broad language spectrum, including formal to informal styles and objective to subjective language, among others, and (3) propose language models inspired by the Logos Model for enhanced transparency, traceability, and customization. The Logos Model is at the core of the best-documented pioneer commercial machine translation system called Logos, which preludes the history of Generative Artificial Intelligence. The open-source version, OpenLogos, offers valuable resources for machine translation, paraphrasing, abstractive summarization, and various other tasks related to Natural Language Generation, and it serves as an excellent resource for training researchers across multiple fields associated with Artificial Intelligence. The OpenLogos resources were made publicly available through the Multi3Generation COST Action (CA18231).

Multi-criteria optimization calculations to improve the efficiency of an agricultural moving power units in fleet

BACKGROUND: Due to the fact that the development of one target function for separately assessing each agricultural machine used in the fleet seems insufficient to fully assess the efficiency of agricultural moving power units in fleet, a multi-criteria formulation of the problem was applied using the previously developed hardware and software complex MOVI 1.4 and the development of computer programs to carry out their optimization calculations of the efficiency of the agricultural moving power units in fleet the example of a specific farm. AIMS: The goal of the work is to develop software tools for carrying out multi-criteria optimization calculations in order to increase the efficiency of agricultural moving power units in fleet in the context of digitalization of agriculture. METHODS: The analysis is based on the collection of scientific publications, scientific articles and other sources of information on the development of research and experimental design work on the creation of intelligent transport and technical means and the improvement of methodological and software for multi-criteria optimization calculations of the efficiency of the agricultural moving power units in fleet. Methods of scientific generalization and statistical processing of available information and analytical materials from domestic and foreign sources were also used. RESULTS: To calculate the values of quality criteria for the MES fleet, an interface subprogram for MOVI 1.4 was developed in the Pascal programming language, adequate for implementing calculations based on mathematical models of criteria at trial points. After introducing functional and criterion restrictions, an admissible set and a set of Pareto-optimal solutions were constructed, and correlation graphs between quality criteria were constructed. The scientific novelty lies in the development of computer programs for multi-criteria justification of the efficiency of the agricultural moving power units in fleet. CONCLUSION: The practical value lies in the fact that the developed mathematical and software tools can be successfully used for multi-criteria optimization of the efficiency of an agricultural moving power units in fleet using the example of specific agricultural organizations, regions or the country as a whole.

Llamol: a dynamic multi-conditional generative transformer for de novo molecular design

Generative models have demonstrated substantial promise in Natural Language Processing (NLP) and have found application in designing molecules, as seen in General Pretrained Transformer (GPT) models. In our efforts to develop such a tool for exploring the organic chemical space in search of potentially electro-active compounds, we present Llamol, a single novel generative transformer model based on the Llama 2 architecture, which was trained on a 12.5M superset of organic compounds drawn from diverse public sources. To allow for a maximum flexibility in usage and robustness in view of potentially incomplete data, we introduce Stochastic Context Learning (SCL) as a new training procedure. We demonstrate that the resulting model adeptly handles single- and multi-conditional organic molecule generation with up to four conditions, yet more are possible. The model generates valid molecular structures in SMILES notation while flexibly incorporating three numerical and/or one token sequence into the generative process, just as requested. The generated compounds are very satisfactory in all scenarios tested. In detail, we showcase the model’s capability to utilize token sequences for conditioning, either individually or in combination with numerical properties, making Llamol a potent tool for de novo molecule design, easily expandable with new properties.Scientific contributionWe developed a novel generative transformer model, Llamol, based on the Llama 2 architecture that was trained on a diverse set of 12.5 M organic compounds. It introduces Stochastic Context Learning (SCL) as a new training procedure, allowing for flexible and robust generation of valid organic molecules with up to multiple conditions that can be combined in various ways, making it a potent tool for de novo molecular design.