Francesco Alesiani, Jonathan Warrell, Tanja Bien
et al.
We propose LOGDIFF (Logical Guidance for the Exact Composition of Diffusion Models), a guidance framework for diffusion models that enables principled constrained generation with complex logical expressions at inference time. We study when exact score-based guidance for complex logical formulas can be obtained from guidance signals associated with atomic properties. First, we derive an exact Boolean calculus that provides a sufficient condition for exact logical guidance. Specifically, if a formula admits a circuit representation in which conjunctions combine conditionally independent subformulas and disjunctions combine subformulas that are either conditionally independent or mutually exclusive, exact logical guidance is achievable. In this case, the guidance signal can be computed exactly from atomic scores and posterior probabilities using an efficient recursive algorithm. Moreover, we show that, for commonly encountered classes of distributions, any desired Boolean formula is compilable into such a circuit representation. Second, by combining atomic guidance scores with posterior probability estimates, we introduce a hybrid guidance approach that bridges classifier guidance and classifier-free guidance, applicable to both compositional logical guidance and standard conditional generation. We demonstrate the effectiveness of our framework on multiple image and protein structure generation tasks.
Classifier-Free Guidance (CFG) serves as the de facto control mechanism for conditional diffusion, yet high guidance scales notoriously induce oversaturation, texture artifacts, and structural collapse. We attribute this failure to a geometric mismatch: standard CFG performs Euclidean extrapolation in ambient space, inadvertently driving sampling trajectories off the high-density data manifold. To resolve this, we present Manifold-Optimal Guidance (MOG), a framework that reformulates guidance as a local optimal control problem. MOG yields a closed-form, geometry-aware Riemannian update that corrects off-manifold drift without requiring retraining. Leveraging this perspective, we further introduce Auto-MOG, a dynamic energy-balancing schedule that adaptively calibrates guidance strength, effectively eliminating the need for manual hyperparameter tuning. Extensive validation demonstrates that MOG yields superior fidelity and alignment compared to baselines, with virtually no added computational overhead.
Daniela Morera-Ulate, Dana Angélica Navarro-Bonilla, José Antonio García-Martínez
Objective: This article analyzes the professional development in Information and Communication Technologies (ICT) and the institutional support for guidance counselors in public schools in Costa Rica.
Methodology: The methodology is quantitative, with an ex post facto, cross-sectional design and descriptive scope. The sample (n=328) comprises guidance professionals from the Ministry of Public Education, who completed a questionnaire.
Results: The main findings reveal a critical perception of institutional support, particularly in aspects related to support for the exchange of technological experiences and attention to professional development needs. Although most of the sample reports receiving ICT training, the most valued modalities are those that promote collaborative learning and flexible access to resources. In contrast, less contextualized options, such as visits to other schools or short courses, are considered less effective.
Conclusions: The importance of collaborative learning and access to updated resources for efficient training is highlighted. However, the lack of contextualization in some modalities and the limited institutional support hinder the integration of ICTs in guidance counseling. The design of practical and relevant programs is recommended, with greater involvement of the educational institutions in identifying the technological needs of the guidance counseling community.
Guidance provides a simple and effective framework for posterior sampling by steering the generation process towards the desired distribution. When modeling discrete data, existing approaches mostly focus on guidance with the first-order Taylor approximation to improve the sampling efficiency. However, such an approximation is inappropriate in discrete state spaces since the approximation error could be large. A novel guidance framework for discrete data is proposed to address this problem: We derive the exact transition rate for the desired distribution given a learned discrete flow matching model, leading to guidance that only requires a single forward pass in each sampling step, significantly improving efficiency. This unified novel framework is general enough, encompassing existing guidance methods as special cases, and it can also be seamlessly applied to the masked diffusion model. We demonstrate the effectiveness of our proposed guidance on energy-guided simulations and preference alignment on text-to-image generation and multimodal understanding tasks. The code is available through https://github.com/WanZhengyan/Discrete-Guidance-Matching/tree/main.
Mark A. Kramer, Aanchal Mathur, Caroline E. Adams
et al.
This paper explores the use of large language models (LLMs) to assist in the development of new disease modules for Synthea, an open-source synthetic health data generator. Incorporating LLMs into the module development process has the potential to reduce development time, reduce required expertise, expand model diversity, and improve the overall quality of synthetic patient data. We demonstrate four ways that LLMs can support Synthea module creation: generating a disease profile, generating a disease module from a disease profile, evaluating an existing Synthea module, and refining an existing module. We introduce the concept of progressive refinement, which involves iteratively evaluating the LLM-generated module by checking its syntactic correctness and clinical accuracy, and then using that information to modify the module. While the use of LLMs in this context shows promise, we also acknowledge the challenges and limitations, such as the need for human oversight, the importance of rigorous testing and validation, and the potential for inaccuracies in LLM-generated content. The paper concludes with recommendations for future research and development to fully realize the potential of LLM-aided synthetic data creation.
The progress in generative AI has fueled AI-powered tools like co-pilots and assistants to provision better guidance, particularly during data analysis. However, research on guidance has not yet examined the perceived efficacy of the source from which guidance is offered and the impact of this source on the user's perception and usage of guidance. We ask whether users perceive all guidance sources as equal, with particular interest in three sources: (i) AI, (ii) human expert, and (iii) a group of human analysts. As a benchmark, we consider a fourth source, (iv) unattributed guidance, where guidance is provided without attribution to any source, enabling isolation of and comparison with the effects of source-specific guidance. We design a five-condition between-subjects study, with one condition for each of the four guidance sources and an additional (v) no-guidance condition, which serves as a baseline to evaluate the influence of any kind of guidance. We situate our study in a custom data preparation and analysis tool wherein we task users to select relevant attributes from an unfamiliar dataset to inform a business report. Depending on the assigned condition, users can request guidance, which the system then provides in the form of attribute suggestions. To ensure internal validity, we control for the quality of guidance across source-conditions. Through several metrics of usage and perception, we statistically test five preregistered hypotheses and report on additional analysis. We find that the source of guidance matters to users, but not in a manner that matches received wisdom. For instance, users utilize guidance differently at various stages of analysis, including expressing varying levels of regret, despite receiving guidance of similar quality. Notably, users in the AI condition reported both higher post-task benefit and regret.
An employer contracts with a worker to incentivize efforts whose productivity depends on ability; the worker then enters a market that pays him contingent on ability evaluation. With non-additive monitoring technology, the interdependence between market expectations and worker efforts can lead to multiple equilibria (contrasting Holmstrom (1982/1999); Gibbons and Murphy (1992)). We identify a sufficient and necessary criterion for the employer to face such strategic uncertainty--one linked to skill-effort complementarity, a pervasive feature of labor markets. To fully implement work, the employer optimally creates private wage discrimination to iteratively eliminate pessimistic market expectations and low worker efforts. Our result suggests that present contractual privacy, employers' coordination motives generate within-group pay inequality. The comparative statics further explain several stylized facts about residual wage dispersion.
در سالهای اخیر مدلهای کسبوکار چرخشی مورد توجه بسیاری از حوزههای کسبوکار قرار گرفته است. حوزه کشاورزی و غذایی نیز با چالشهای جدیدی بهمنظور کاهش تلفات و ضایعات مواد غذایی در طول زنجیره تامین خود روبرو میباشد. در این زمینه، پذیرش و کاربست مدلهای کسبوکار چرخشی، فرصتهایی را برای ارتقای انتقال به سمت عملکرد پایدارتر و آیندهای از نظر اقتصادی مناسبتر فراهم میکند. از اینرو پژوهش حاضر با رویکرد کیفی و با روش فراترکیب به شناسایی مزایای کاربست مدلهای کسبوکار چرخشی در حوزه کشاورزی- غذایی پرداخته است. در همین راستا، تعداد 57 مطالعه مرتبط ارزیابی و درنهایت، پس از بررسی 29 مقاله انتخاب شد و با استفاده از الگوی هفت مرحلهای سندلوسکی و باروسو، در دو مرحله کدگذاری باز و کدگذاری محوری تحلیل گردیدند. به این ترتیب مهمترین مزایای شناسایی شده کاربست مدلهای کسبوکار چرخشی در حوزه کشاورزی-غذایی در پنج طبقه اصلی اقتصادی، زیستمحیطی، زنجیره ارزش، بازاریابی و اجتماعی دستهبندی شدند. در پایان، با توجه به گزارههای پژوهش، پیشنهادهایی ارائه گردید. در یک نمونه مهم، سیاستگذاران و دستاندرکاران حوزه کشاورزی-غذایی بایستی آگاهی جامعه (از طریق فرهنگسازی بهوسیله صداوسیما و مدارس) و کشاورزان (از طریق برگزاری دورهها و کمک به ارزشگذاری مجدد محصول و محصولات جانبی و برچسبزنی آنها) را افزایش دهند. بهعلاوه ایجاد تسهیلات و مشوقهای مالی نیز میتوانند در فراهم نمودن مقدمات اجرا و موفقیت مدلهای کسبوکار چرخشی در حوزه کشاورزی-غذایی اثرگذار باشند.
Vocational guidance. Career development, Agriculture (General)
In this practice article, that builds on our experiences from a Norwegian context, we argue for a shift in the traditional performance interview towards what we call career interviews. Underpinned by relevant theory, we show how a reconstruction of the traditional performance interview into what we call career interview may restore and maintain the balance in the game of exchange between employer and employee, and secure knowledge sharing processes to the benefit of both the organisation and the individual. Tools from the Norwegian Quality framework for career guidance are used to show the practical utilisation of such a shift. Abstrakt Med utgangspunkt i en norsk kontekst argumenteres det i denne praksisartikkelen for hvordan et skifte i den tradisjonelle medarbeidersamtalen til det vi kaller karrieresamtaler kan medvirke til en balansert prosess hvor både arbeidsgiver og organisasjoners behov for videreutvikling ivaretas. Samtalestartere, et verktøy hentet fra nasjonalt kvalitetsrammeverk for karriereveiledning, brukes som eksempel på hvordan de foreslåtte karrieresamtalene kan gjennomføres. Nøkkelord: Medarbeidersamtale; karrierelæring; kunnskapsutvikling i arbeidslivet; organisasjonsutvikling
This paper studies AI persuasion by distinguishing between two reasons for disagreement: attention differences, where the AI detects features the decision-maker missed, and comprehension differences, where the AI and the decision-maker interpret observed features differently. We show that AI is more effective in persuading the decision-maker when the disagreement is due to attention differences rather than comprehension differences. We also show that the AI's interpretability shapes how the decision-maker attributes the sources of disagreement and, in turn, whether they follow the AI's recommendation. Our main result is that making AI uninterpretable can actually enhance persuasion and, in the presence of career concerns, improve decision accuracy.
Diffusion models excel in generating high-quality images. However, current diffusion models struggle to produce reliable images without guidance methods, such as classifier-free guidance (CFG). Are guidance methods truly necessary? Observing that noise obtained via diffusion inversion can reconstruct high-quality images without guidance, we focus on the initial noise of the denoising pipeline. By mapping Gaussian noise to `guidance-free noise', we uncover that small low-magnitude low-frequency components significantly enhance the denoising process, removing the need for guidance and thus improving both inference throughput and memory. Expanding on this, we propose \ours, a novel method that replaces guidance methods with a single refinement of the initial noise. This refined noise enables high-quality image generation without guidance, within the same diffusion pipeline. Our noise-refining model leverages efficient noise-space learning, achieving rapid convergence and strong performance with just 50K text-image pairs. We validate its effectiveness across diverse metrics and analyze how refined noise can eliminate the need for guidance. See our project page: https://cvlab-kaist.github.io/NoiseRefine/.
Purwanto, Hapsari Peni, Agustin Tjahyaningtijas
et al.
Clustering the distribution of student graduates is an approach used to analyze and understand the success of Vocational High School education programs in preparing graduates to enter the workforce or start their own businesses. The purpose of clustering is to evaluate the effectiveness of educational programs, identify entrepreneurial potential, formulate career planning, and develop entrepreneurial skills, all contributing to the fulfilment of Sustainable Development Goals (SDGs) related to quality education (SDG 4), decent work and economic growth (SDG 8), and industry, innovation, and infrastructure (SDG 9). Through this clustering, schools can evaluate the extent to which quality high school graduates achieve career or entrepreneurial success, supporting the objectives of SDG 4. This information helps in designing educational programs that are more in line with the needs of the job market, providing better career guidance to students, and promoting entrepreneurial skills among high school students, contributing to SDG 4 and SDG 8. Clustering the distribution of vocational high school students by working, continuing, and entrepreneurial status plays an important role in strengthening the link between education and the world of work, aligning with the aims of SDG 4 and SDG 8. Self-Organizing Map (SOM), as an Artificial Neural Network, assists in data clustering or mapping tasks, aiding in the discovery of patterns and trends within the vocational high school graduate population. The result of clustering using Z-Score and Min-Max normalization techniques is 5.31% and 3.98%, respectively, providing insights into the career and entrepreneurship trends and patterns of vocational high school students. This valuable information can be used for the development of educational programs, career guidance initiatives, and improved alignment between education and the needs of the world of work, ultimately contributing to the realization of SDGs 4, 8, and 9.
Problem statement . Every year, immersive equipment becomes more accessible, and the material and technical base of educational organizations is filled with new innovative devices. An urgent scientific task is the development of methods and tools for the implementation and usage of immersive devices in the educational process. Methodology. The authors describe approaches to the informatization of education based on the model of the center of immersive technologies (CIT), which contributes to the solution of managerial, methodical, pedagogical and technical problems associated with the application of immersive technologies in the educational and career guidance processes of educational organizations. The main functions of the CIT model are divided into three components: training students of secondary vocational education (SVE) in working professions, the realization of short programs for advanced training and assessment of qualifications, career guidance of schoolchildren. The key users of the CIT model are SVE students, schoolchildren and their parents, the management team of the educational organizations, teachers. The CIT model can be implemented in any organization that has immersive equipment (VR/AR) and human resources. Results. The implementation of immersive simulators in teaching students in professional programs, educational practice, as well as in organizing and conducting professional excellence championships demonstrates approximately equal effectiveness in comparison with traditional teaching methods, except for the cases in which it is important to develop tactile skills in working with certain equipment. Using the example of the execution of VR simulators in the teaching of the subject “Technology” in the school course, the advantage of learning on real machines is demonstrated. It was discovered that the maximum pedagogical effect is achieved by adding VR simulators to the training course and by using a combined approach. VR simulators are necessary for work in the absence of workshops and serve as an important didactic propaedeutic tool when they are available. Conclusion. The domestic and foreign experience of using immersive technologies in educational and outreach activities is analyzed, the development projects of the Russian Federation in the field of immersive technologies are considered. The relevance, theoretical and practical significance of the development and implementation of the CIT model are substantiated. The results of approbation in educational institutions of secondary vocational and basic general education are presented.
سودابه ابراهیمی, عادل زاهد بابلان, مهدی معینی کیا
et al.
پژوهش حاضر با هدف ارتباطیابی نقش شادابی سازمانی و توانمندسازی روانشناختی در مدیریت ترومای سازمانی نو معلمان با رویکرد شبکه عصبی مصنوعی انجام گرفت. جامعه آماری این مطالعه شامل نو معلمان استان گیلان، البرز و تهران با جمعیت 15000 نفر بود. روش نمونهگیری از نوع خوشهای چند مرحلهای بود. حجم نمونه با توجه به مدل کرجسی- مورگان 375 نفر در نظر گرفته شد. برای جمعآوری دادهها از پرسشنامه شادابی سازمانی کرولف ، پرسشنامه توانمندسازی روانشناختی اسپرتیز و پرسشنامه مدیریت ترومای سازمانی محقق ساخته استفاده شد. روایی ابزارها با نظر استادان علوم تربیتی و روانشناسی تأیید شد. دادهها با شبکه عصبی مصنوعی پرسپترون چند لایه (MLP) در نرم افزار Spss.25 تحلیل شد. نتایج نشان داد ارتباطیابی شادابی سازمانی و توانمندسازی روانشناختی بر مدیریت ترومای سازمانی دارای یک لایه ورودی با 13 گره و یک لایه پنهان با 4 گره است و شبکه عصبی مصنوعی به خوبی قادر است پرشها و روند مدیریت ترومای سازمانی را از روی شادابی سازمانی و توانمندسازی روانشناختی پیشبینی کند. یافتهها نشان میدهد که احساس مشارکت با دیگران بیشترین تأثیر و معنادار بودن کار کمترین تأثیر ضریب اهمیت را در بر آورد مدیریت ترومای سازمانی در بر میگیرد. با توجه به بررسیها، بحرانهای روانشناختی در سازمانها به یک تجربه رایج برای مدیران در سراسر جهان تبدیل شده است. اگر به درستی مدیریت نشود به نظر میرسد که تجربه آسیب جمعی در قالب بسیاری از عوامل سازمانی بروز میدهد که منجر به رفتار ناکارآمد در سازمان میشود.
کسبوکارهای خانگی آنلاین با وجود اهمیت فراوان از نظر اشتغالزایی و کارآفرینی، با چالشها و مشکلات فراوانی مواجه هستند که میتواند بقا و توسعه آنها را به خطر بیندازد. هدف پژوهش حاضر، تبیین نقش تبلیغات دهانبهدهان الکترونیک (تدبدا) بر اعتمادسازی و پیشبرد فروش کسبوکارهای خانگی آنلاین است. این پژوهش کاربردی بوده، از حیث هدف توصیفی است. اعضای نمونهی آماری این پژوهش، ۳۸۶ نفر از مصرفکنندگان ساکن مازندران با تجربه حداقل یکبار خرید محصولات کسبوکارهای خانگی در بستر رسانههای اجتماعی بودند که با استفاده از روش نمونهگیری غیر تصادفی در دسترس و از طریق پرسشنامههای تدبدا، پیشبرد فروش و اعتماد آنلاین مورد بررسی قرار گرفتند. پایایی ابزار تحقیق به وسیله بررسی سازگاری درونی و روایی آن از طریق روایی همگرا و روایی واگرا مورد سنجش و تائید قرار گرفت. جهت بررسی فرضیات تحقیق از مدل معادلات ساختاری استفاده شد. نتایج نشان داد در سطح اطمینان پنج درصد اعتماد آنلاین تأثیر مستقیم، مثبت و معناداری بر پیشبرد فروش دارد و دو بعد محتوای تدبدا و جاذبه مثبت تدبدا به طور مثبت و بعد جاذبه منفی تدبدا به طور منفی پیشبرد فروش را تحت تأثیر قرار میدهند. همچنین تدبدا بهعنوان متغیر تعدیلگر میتواند اثر متغیر اعتماد آنلاین بر پیشبرد فروش را افزایش دهد. یافتهها نشان داد پیشبرد فروش وابستگی زیادی به اعتماد آنلاین دارد و این رابطه نیز به وسیله تدبدا به صورت مثبت یا منفی تحت تأثیر قرار میگیرد.
Vocational guidance. Career development, Agriculture (General)
PROs developed de novo, using the FDA guidance may involve structured patient interviews or focus groups. Qualitative Research is a methodology for eliciting and coding interviews and produces concepts or themes. These concepts are used to develop items in a PRO for use as an endpoint in Clinical trials. A convention in the field is that interviews and code/concept elicitation are considered complete when subsequent interviews produces "no new concepts" -termed "saturation". FDA reviewers frequently challenge PRO developers whether there are sufficient patient interviews to confirm that saturation is achieved after occurrence of zero new concepts. Several authors have reported that concrete criteria are need for confirming that saturation is achieved (Francis 2010, Mason 2010, Marshall 2013). I provide statistical methodology for confirming saturation, suitable for review by a regulatory authority. Type I error for saturation, may occur if further interviews elicited more concepts after first occurrence of saturation. I use published data set on code elicitation (Guest, 2006) to demonstrate that saturation may occur more than once in a sequence of interviews. I provide a statistical definition for saturation in qualitative research, that addresses regulatory concerns for PRO's developed for use as a clinical trial endpoint in a regulatory submission.
Popular guidance for denoising diffusion probabilistic model (DDPM) linearly combines distinct conditional models together to provide enhanced control over samples. However, this approach overlooks nonlinear effects that become significant when guidance scale is large. To address this issue, we propose characteristic guidance, a guidance method that provides first-principle non-linear correction for classifier-free guidance. Such correction forces the guided DDPMs to respect the Fokker-Planck (FP) equation of diffusion process, in a way that is training-free and compatible with existing sampling methods. Experiments show that characteristic guidance enhances semantic characteristics of prompts and mitigate irregularities in image generation, proving effective in diverse applications ranging from simulating magnet phase transitions to latent space sampling.
This paper analyzes the professional bottlenecks limiting the career development of young teachers in vocational colleges, including insufficient teaching skills, limited research capacity, and restricted professional development and career progression. Targeted recommendations involve fostering young teachers’ teaching skills and research capabilities through mentorship, training, research participation and skills competition guidance. Addressing these bottlenecks effectively develops young teachers’ competencies, contributing innovative ideas and practices to improve vocational education quality, cementing their role as the next generation of vocational educators.
The present study examines the preparation of students for future career development in the country and abroad. The note is paid to the professions that the Vocational High School of Tourism – Ruse offers for the implementation of early career guidance for all students, through additional training courses, work on projects, meetings with foreign specialists, exchange of experience with schools from other countries and competitions regional and national in nature. In the research, the performances of the High School students in the field of tourism and their planned career development are observed.