Sharif Digital Repository

The advancement of large language models (LLMs), their ability to produce eloquent and fluent content, and their vast knowledge have resulted in their usage in various tasks and applications. Despite generating fluent content, this content can contain fabricated or false information. This problem is known as hallucination and has reduced the confidence in the output of LLMs. In this work, we have used Natural Language Inference to train classifiers for hallucination detection to tackle SemEval-2024 Task 6-SHROOM (Mickus et al., 2024) which is defined in three sub-tasks: Paraphrase Generation, Machine Translation, and Definition Modeling. We have also conducted experiments on LLMs to evaluate... 

Inspired by an upward stochastic walking idea, a new ensemble pruning method called simulated quenching walking (SQWALKING) is developed in this paper. The rationale behind this method is to give values to stochastic movements as well as to accept unvalued solutions during the investigation of search spaces. SQWALKING incorporates simulated quenching and forward selection methods to choose the models through the ensemble using probabilistic steps. Two versions of SQWALKING are introduced based on two different evaluation measures; SQWALKINGA that is based on an accuracy measure and SQWALKINGH that is based on a human-like foresight measure. The main objective is to construct a proper... 

The goal and dream of the artificial intelligence field have long been the development of intelligent systems or agents that mimic human behavior and thinking. Creativity is an essential trait in humans that is closely related to lateral thinking. The remarkable advancements in Language Models have led to extensive research on question-answering and explicit and implicit reasoning involving vertical thinking. However, there is an increasing need to shift focus towards research and development of models that can think laterally. One must step outside the traditional frame of commonsense concepts in lateral thinking to conclude. Task 9 of SemEval-2024 is Brainteaser (Jiang et al., 2024), which... 

The front double wishbone suspension parameters are related as a multi-body model. By replacing the new suspension and steering hard points, the performance of the suspension system is studied using CATIA and MSC ADAMS software for modeling and analyzing, respectively. In this case two tests were done, steady state circular test in ADAMS/View and parallel wheel travel in ADAMS/Car. The strut position is directly related to Kingpin inclination angle and scrub radius. The increase the amount of these two parameters has an effect on steering performance in low speed and negative displacement, such as camber, caster and Kingpin inclination angles in negative displacement of wheels. Moreover, the... 

Semantic Textual Relatedness holds significant relevance in Natural Language Processing, finding applications across various domains. Traditionally, approaches to STR have relied on knowledge-based and statistical methods. However, with the emergence of Large Language Models, there has been a paradigm shift, ushering in new methodologies. In this paper, we delve into the investigation of sentence-level STR within Track A (Supervised) by leveraging fine-tuning techniques on the RoBERTa transformer. Our study focuses on assessing the efficacy of this approach across different languages. Notably, our findings indicate promising advancements in STR performance, particularly in Latin languages.... 

Novelty Detection (ND) plays a crucial role in machine learning by identifying new or unseen data during model inference. This capability is especially important for the safe and reliable operation of automated systems. Despite advances in this field, existing techniques often fail to maintain their performance when subject to adversarial attacks. Our research addresses this gap by marrying the merits of nearest-neighbor algorithms with robust features obtained from models pretrained on ImageNet. We focus on enhancing the robustness and performance of ND algorithms. Experimental results demonstrate that our approach significantly outperforms current state-of-the-art methods across various... 

Language models, particularly generative models, are susceptible to hallucinations, generating outputs that contradict factual knowledge or the source text. This study explores methods for detecting hallucinations in three SemEval-2024 Task 6 tasks: Machine Translation, Definition Modeling, and Paraphrase Generation. We evaluate two methods: semantic similarity between the generated text and factual references, and an ensemble of language models that judge each other's outputs. Our results show that semantic similarity achieves moderate accuracy and correlation scores in trial data, while the ensemble method offers insights into the complexities of hallucination detection but falls short of... 

Detecting Machine-Generated Text (MGT) has emerged as a significant area of study within Natural Language Processing. While language models generate text, they often leave discernible traces, which can be scrutinized using either traditional feature-based methods or more advanced neural language models. In this research, we explore the effectiveness of fine-tuning a RoBERTa-base transformer, a powerful neural architecture, to address MGT detection as a binary classification task. Focusing specifically on Subtask A (Monolingual - English) within the SemEval-2024 competition framework, our proposed system achieves an accuracy of 78.9% on the test dataset, positioning us at 57th among... 

Conventional implementation of multi-dimensional wavelet transform (e.g. 3-D wavelet) requires whether a high amount of "in access" memory or a continual access to slow memory of a processor which makes it infeasible for most applications. In this paper, we proposed a novel algorithm for computation of an n-D discrete wavelet transform (DWT) based on lifting scheme. In addition to benefits of lifting scheme (which causes a major reduction in computational complexity and performs the total computations in time domain), our real-time approach computes the coefficients for all kinds of 1st and 2 nd generation wavelets with short delay and optimized utilization of the slow and fast memories of a... 

The heat un-balance and the immiscibility of the Cu and Fe alloy system are the key challenges during Cu/Fe resistance spot welding. Utilising electrodes of different geometries enabled the formation of a mixed Cu–Fe fusion zone. It is identified that despite the immiscibility of Cu and Fe in the solid-state, metastable liquid phase separation enables a metallurgical joining between Cu and Fe. The liquid phase spinodal decomposition led to the formation of an ultra-fine interconnected network of Cu-rich and Fe-rich phases in the fusion zone. The higher hardness of the fusion zone compared to the base metals, the susceptibility to solidification cracking, and the electrode indentation in the...