Middle East Technical University Research Information System
Not a member yet
104897 research outputs found
Sort by
Ductile fracture prediction for flow forming of Inconel 718 with experimental validation and finite element simulations
No full textFlow forming is an advanced metal forming technique that allows the production of thin-walled, axisymmetric components with high dimensional accuracy and mechanical integrity. However, because preforms are subjected to complex stress states and intense plastic deformation during forming, geometric distortions and ductile fractures can occur, especially at high reduction ratios. This study provides a detailed analysis of widely used uncoupled ductile damage models for predicting fracture behavior during the flow forming of Inconel 718 alloy. Fifteen damage criteria, including both single- and multi-parameter damage models, are calibrated using tensile tests for four different geometries representing varying stress states. The models are implemented using a user-defined subroutine (VUSDFLD) in Abaqus/Explicit. The calibrated models are applied to both tensile tests and the flow forming process, with the results validated against experimental data. The findings indicate that the Ayada model provides more accurate damage predictions across all reduction ratios compared to other models, making it particularly suitable for the flow forming process. Furthermore, the influence of process parameters such as feed rate, revolution speed, feed ratio, and roller offset on formability and fracture initiation is investigated. The results underscore the crucial importance of selecting suitable process parameters and optimizing the forming process
A Novel Framework for Next Word Prediction Using Long-Short Term Memory Networks
No full textNatural Language Processing (NLP) has become a cornerstone in various fields, revolutionizing how machines interpret and process human language. Among its diverse applications, next-word prediction emerges as a highly practical and impactful example of generative AI. This research focuses on the use of Long Short-Term Memory (LSTM) models—an innovative class of Recurrent Neural Network (RNN)—for predictive text generation. LSTMs excel in capturing sequential and contextual information, making them ideal for language tasks. While transformer models dominate accuracy benchmarks, this work addresses the critical need for efficient alternatives in resource-constrained deployment scenarios. This study presents a novel LSTM-based framework enhanced with hybrid architecture and advanced regularization techniques, trained on a carefully curated dataset of 15,000 English sentences. The proposed model achieves superior performance with 84.2% training accuracy, 79.6% test accuracy, and a perplexity score of 2.41, significantly outperforming traditional approaches. The methodology addresses overfitting through dropout regularization, batch normalization, and adaptive learning rate strategies while effectively capturing long-term contextual dependencies. This research contributes to the advancement of neural language modeling by providing a robust framework that bridges the gap between computational efficiency and prediction accuracy in real-world NLP applications.</p
Solution Approaches for the Dynamic Naval Air Defense Planning Problem
No full textThe naval air defense planning (NADP) problem entails the defense of a naval fleet against aerial threats. This complex and dynamic problem requires real-time decision-making and adaptation to evolving warfare environment. While our previous work addressed the static NADP problem by proposing a mathematical model and heuristic solutions for sensor allocation, engagement scheduling, and ship routing, this study extends to the dynamic NADP problem. Unlike the static version, which assumes complete knowledge of future threats, the dynamic NADP problem requires continuous updates and real-time adjustments to decisions as new threats emerge and situational parameters change. We present modifications in the mathematical formulation, which is based on a mixed-integer nonlinear programming (MINLP) model, alongside a comprehensive simulation structure. We employ heuristic solution approaches that utilize a combination of a genetic algorithm, construction of an engagement graph to solve the shortest path problem, and dynamic programming (DP) techniques. Computational experiments are conducted to evaluate the effectiveness of these methods in addressing the dynamic NADP problem. The study also explores machine learning models for threat prioritization, offering innovative solutions to the challenges posed by dynamic naval air defense scenarios
Increasing and other subsequence problems for random interval sequences
No full textVarious relations for comparison of intervals of real numbers are introduced, and the expected length of the corresponding longest increasing subsequence is analyzed. When intervals are randomly generated by taking the minimum and maximum of two independent uniform random variables, we prove that the expected length of the longest increasing subsequence grows on the order of n3. We also investigate the asymptotic behavior of the expected length under alternative comparison relations and random interval models. Discussions on other subsequence problems for interval sequences are included
Monitoring power in context: A hermeneutic exploration of relational dynamics in counseling and psychotherapy supervision
No full textObjective: This study explored relational dynamics within a supervisory relationship, focusing on power, multicultural considerations, and parallel processes. Method: Guided by Gadamerian philosophical hermeneutics, the exploration relied on an iterative, reflexive engagement with eight recorded supervision sessions to interpret how these dynamics unfolded across time. Reflexive thematic analysis was used within a hermeneutic framework to identify relational patterns, and the Shedler–Westen Assessment Procedure served as an additional interpretive layer, providing the supervisor’s professional perspective on supervisee personality tendencies. Results: Findings highlighted consistently fluctuating power negotiations, relationally embedded meanings shaping safety and disclosure, and tentative parallel processes that reflected patterns described in the supervisee’s site supervision and client work. These dynamics appeared as interpretive layers rather than fixed mechanisms, emerging through the dialogical flow of supervision. Conclusion: The study offers process-level insights into how relational dynamics take shape within lived supervision, illustrating the value of hermeneutic inquiry for understanding supervision as an unfolding interpersonal experience. Implications for supervisors include attending to relational patterns as tentative cues for deeper inquiry
Designing and developing a mobile teacher professional development course for digital game-enhanced language learning
No full textWhile digital games hold promise for language learning, teachers often lack the knowledge and confidence to integrate them into their curriculum. Existing professional development opportunities are scarce and not always accessible. In this study, a mobile teacher professional development course on Digital Game-Enhanced Language Learning was designed, developed, and evaluated through three iterative cycles. The research employed a Type 1 design and development research methodology. Three groups of language instructors teaching English at the tertiary level participated in the iterative cycles. Data collection included course evaluation questionnaires, achievement tests, lesson plan development, and participant interviews. The findings indicated that the mobile teacher professional development course was effective in improving teacher perceptions of the course content, usability, and self-efficacy in integrating digital games into their teaching. The course also led to a significant increase in participants' knowledge of Digital Game-Enhanced Language Learning concepts. Overall, the mobile teacher professional development course has the potential to be a valuable tool for equipping language teachers with the skills and confidence needed to leverage digital games in their classrooms. This is particularly relevant considering the growing need for accessible and flexible professional development opportunities, such as during a global crisis
A Stanislavskian Framework for Architectural Design: Creating Lively Spaces Through ‘Given Circumstances’ and ‘Magic if’
No full textThis study investigates the conceptual and operative parallels between Stanislavsky’s methods and the architectural design process based on architecture and theatre’s shared engagement with human life: architecture designs spaces for life, and theatre represents life situations. The study introduces vitalness as a term to describe architecture and theatre’s shared capacity to contain a dynamic life flow within a multi-layered milieu; it articulates the term through two components–the human being and its phenomenal environment–and examines how it is generated in these disciplines. Arguing that architects think like actors in the design process to envision vitalness, the study proposes that Stanislavsky’s acting methods can benefit architects’ imaginative reasoning. Employing a theoretical and critical analytical methodology, the study surveys the literature to identify architecture and theatre’s common ground and proposes a conceptual framework of vitalness which underscores the two Stanislavskian concepts aligned with the components of vitalness: given circumstances and magic if. The study lastly analyses the creative production processes with a focus on the parallels between Stanislavsky’s method and the architectural design process. This analysis proposes a Stanislavskian lens to spatial design and serves as a systematic standpoint to creativity and imagination, facilitating human-based thinking across disciplinary boundaries
Intrinsic dimensionality as a model-free measure of class imbalance
No full textImbalance in classification tasks is commonly quantified by the cardinalities of examples across classes. This, however, disregards the presence of redundant examples and inherent differences in the learning difficulties of classes. Alternatively, one can use complex measures such as training loss and uncertainty, which, however, depend on training a machine learning model. Our paper proposes using data Intrinsic Dimensionality (ID) as an easy-to-compute, model-free measure of imbalance that can be seamlessly incorporated into various imbalance mitigation methods. Our results across five different datasets with a diverse range of imbalance ratios show that ID consistently outperforms cardinality-based re-weighting and re-sampling techniques used in the literature. Moreover, we show that combining ID with cardinality can further improve performance. Our code and models are available at https://github.com/cagries/IDIM