Papers by Amir Hossein Rahimi
Context: User intent modeling is a crucial process in Natural Language Processing that aims to id... more Context: User intent modeling is a crucial process in Natural Language Processing that aims to identify the underlying purpose behind a user’s request, enabling personalized responses. With a vast array of approaches introduced in the literature (over 13,000 papers in the last decade), understanding the related concepts and commonly used models in AI-based systems is essential. Method: We conducted a systematic literature review to gather data on models typically employed in designing conversational recommender systems. From the collected data, we developed a decision model to assist researchers in selecting the most suitable models for their systems. Additionally, we performed two case studies to evaluate the effectiveness of our proposed decision model. Results: Our study analyzed 59 distinct models and identified 74 commonly used features. We provided insights into potential model combinations, trends in model selection, quality concerns, evaluation measures, and frequently used ...
Journal of Research & Health
Background: Coronavirus disease 2019 is a new viral disease and in a short period, it has affecte... more Background: Coronavirus disease 2019 is a new viral disease and in a short period, it has affected the world in various economic, social, and health aspects. This disease had a high mortality rate at the time of its occurrence. This article aims to determine the related factors to the survival time of inpatients with COVID-19 in northern Iran. Methods: In this retrospective study, the data of 3480 patients with laboratory confirmation of the virus caused by COVID-19 infection in 17 hospitals covered by Golestan University of Medical Sciences from February 20 to July 20, 2020 were used. For all patients included in the study, until the end of the study, the follow-up has been done through the hospital information system unit. Statistical analysis was performed using R statistical software version 3.6.2 with a survival package. Results: Out of 3480 definitive patients with COVID-19, the average age of the patients was 52.93±19.20 years and 51.1% of them were women. A total of 77.6% of...
arXiv (Cornell University), Mar 15, 2020
Predicting calibrated confidence scores for multi-class deep networks is important for avoiding r... more Predicting calibrated confidence scores for multi-class deep networks is important for avoiding rare but costly mistakes. A common approach is to learn a post-hoc calibration function that transforms the output of the original network into calibrated confidence scores while maintaining the network's accuracy. However, previous post-hoc calibration techniques work only with simple calibration functions, potentially lacking sufficient representation to calibrate the complex function landscape of deep networks. In this work, we aim to learn general post-hoc calibration functions that can preserve the top-k predictions of any deep network. We call this family of functions intra order-preserving functions. We propose a new neural network architecture that represents a class of intra order-preserving functions by combining common neural network components. Additionally, we introduce order-invariant and diagonal sub-families, which can act as regularization for better generalization when the training data size is small. We show the effectiveness of the proposed method across a wide range of datasets and classifiers. Our method outperforms state-of-the-art post-hoc calibration methods, namely temperature scaling and Dirichlet calibration, in several evaluation metrics for the task.
Lecture Notes in Mechanical Engineering
Performance, development and validation of vehicle dynamics applications such as anti-lock brakin... more Performance, development and validation of vehicle dynamics applications such as anti-lock braking systems (ABS) or indirect tyre pressure monitoring systems (iTPMS) rely on a sufficiently accurate consideration of tyre properties such as transient dynamics and tyre kinematics. Previous investigations showed that the tyre rotation has a distinct effect on the vertical tyre stiffness as well as on the three characteristic tyre radii, the unloaded, static and effective (dynamic) tyre radius. Based on the fundamentals of the TMeasy 5 handling tyre model, an enhanced semi-physical modelling approach was developed to consider rotational speed dependent tyre stiffness and tyre radii in an effective and numerically efficient manner. In the present study, a detailed experimental validation is conducted to verify, evaluate and validate the previously identified rotational speed induced effects and the developed enhanced modelling approach. Based on the results of an extensive tyre testing se...
Transactions of the Indian Institute of Metals
2019 IEEE/CVF International Conference on Computer Vision (ICCV)
Given a collection of bags where each bag is a set of images, our goal is to select one image fro... more Given a collection of bags where each bag is a set of images, our goal is to select one image from each bag such that the selected images are from the same object class. We model the selection as an energy minimization problem with unary and pairwise potential functions. Inspired by recent few-shot learning algorithms, we propose an approach to learn the potential functions directly from the data. Furthermore, we propose a fast greedy inference algorithm for energy minimization. We evaluate our approach on few-shot common object recognition as well as object co-localization tasks. Our experiments show that learning the pairwise and unary terms greatly improves the performance of the model over several well-known methods for these tasks. The proposed greedy optimization algorithm achieves performance comparable to state-of-the-art structured inference algorithms while being ∼10 times faster.
Journal of Production and Operations Management, 2019
Supply Chain Management (SCM) is a suitable tool to improve economic, social and environmental pe... more Supply Chain Management (SCM) is a suitable tool to improve economic, social and environmental performance. SCM assessment is an important task for all types of organizations. The DEA method has been widely used to evaluate SCM. By attention supply chain as network data envelopment analysis (DEA) can calculate the efficiency of supply chain with multiple stages. This study examines the efficiency and returns to scale (RTS) of supply chain management of resin manufacturing companies based on network DEA models. We determine returns to scale of resin manufacturing companies as a two-stage process, with crisp and fuzzy data. Fuzzy DEA model is based on approach to measure the efficiency and RTS of supply chain. The proposed models are used to evaluate the efficiency and RTS of supply chain of 27 resin production companies. The six companies were network efficient in the investigation with crisp data, while there are three network efficient companies with fuzzy data. <strong>Int...
2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2022
Detecting novel objects from few examples has become an emerging topic in computer vision recentl... more Detecting novel objects from few examples has become an emerging topic in computer vision recently. However, these methods need fully annotated training images to learn new object categories which limits their applicability in real world scenarios such as field robotics. In this work, we propose a probabilistic multiple instance learning approach for few-shot Common Object Localization (COL) and fewshot Weakly Supervised Object Detection (WSOD). In these tasks, only image-level labels, which are much cheaper to acquire, are available. We find that operating on features extracted from the last layer of a pre-trained Faster-RCNN is more effective compared to previous episodic learning based few-shot COL methods. Our model simultaneously learns the distribution of the novel objects and localizes them via expectation-maximization steps. As a probabilistic model, we employ von Mises-Fisher (vMF) distribution which captures the semantic information better than Gaussian distribution when applied to the pre-trained embedding space. When the novel objects are localized, we utilize them to learn a linear appearance model to detect novel classes in new images. Our extensive experiments show that the proposed method, despite being simple, outperforms strong baselines in few-shot COL and WSOD, as well as largescale WSOD tasks.
Abstract—The purpose of this study was to investigate the relationship among metacognitive strate... more Abstract—The purpose of this study was to investigate the relationship among metacognitive strategy use, motivation and listening test performance of EFL university students. The participants were 82 students majoring in English translation and literature at Allameh Tabataba’i and Shahid Beheshti Universities in Tehran, Iran. Data were collected using three instruments: MALQ (metacognitive awareness listening questionnaire), AMS (academic motivation scale), and the listening section of the TOEFL. After administering the listening section of the TOEFL (pre-test), students filled in the MALQ and AMS. A statistically significant correlation was found between metacognitive strategy use and listening performance, listening performance and intrinsic motivation, as well as metacognitive strategy use and intrinsic, extrinsic motivation. Index Terms—listening, metacognitive strategies, motivation, self-determination theory I.
ArXiv, 2020
Weakly Supervised Object Localization (WSOL) methods have become increasingly popular since they ... more Weakly Supervised Object Localization (WSOL) methods have become increasingly popular since they only require image level labels as opposed to expensive bounding box annotations required by fully supervised algorithms. Typically, a WSOL model is first trained to predict class generic objectness scores on an off-the-shelf fully supervised source dataset and then it is progressively adapted to learn the objects in the weakly supervised target dataset. In this work, we argue that learning only an objectness function is a weak form of knowledge transfer and propose to learn a classwise pairwise similarity function that directly compares two input proposals as well. The combined localization model and the estimated object annotations are jointly learned in an alternating optimization paradigm as is typically done in standard WSOL methods. In contrast to the existing work that learns pairwise similarities, our proposed approach optimizes a unified objective with convergence guarantee and ...
Die Dokumente in HENRY stehen unter der Creative Commons Lizenz CC BY 4.0, sofern keine abweichen... more Die Dokumente in HENRY stehen unter der Creative Commons Lizenz CC BY 4.0, sofern keine abweichenden Nutzungsbedingungen getroffen wurden. Damit ist sowohl die kommerzielle Nutzung als auch das Teilen, die Weiterbearbeitung und Speicherung erlaubt. Das Verwenden und das Bearbeiten stehen unter der Bedingung der Namensnennung. Im Einzelfall kann eine restriktivere Lizenz gelten; dann gelten abweichend von den obigen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Documents in HENRY are made available under the Creative Commons License CC BY 4.0, if no other license is applicable. Under CC BY 4.0 commercial use and sharing, remixing, transforming, and building upon the material of the work is permitted. In some cases a different, more restrictive license may apply; if applicable the terms of the restrictive license will be binding.
Introduction: Multiple sclerosis (MS) is an immune mediated demyelinating disease of the central ... more Introduction: Multiple sclerosis (MS) is an immune mediated demyelinating disease of the central nervous system. The aim of present study was to investigate the effect of eight weeks resistance training on interlukin-17 (IL-17) in women with MS. Material & Methods: Twenty seven women with MS disease in a range of 18-48 year of old and EDSS lower than 4.5 participated in this study as the subject. Subjects were divided into control group (n=13) or training group (n=14) randomly. The training group performed progressive resistance training program, 3 days a week for 8 weeks, whereas control group continued their usual routine activities. Serum level of IL-17 was measured by ELISA kits before and after training. Results: The disability score and IL-17 were significantly decreased from 1.8 to 1.3 and from 716.3 to 601.3 pg/ml respectively in test MS subjects after 8 weeks resistance training. Conclusions: In summary, the results suggest that resistance training has useful anti-inflammat...
arXiv: Other Statistics, 2018
Innovation in healthcare payment and service delivery utilizes high cost, high risk pilots paired... more Innovation in healthcare payment and service delivery utilizes high cost, high risk pilots paired with traditional program evaluations. Decision-makers are unable to reliably forecast the impacts of pilot interventions in this complex system, complicating the feasibility assessment of proposed healthcare models. We developed and validated a Discrete Event Simulation (DES) model of primary care for patients with Diabetes to allow rapid prototyping and assessment of models before pilot implementation. We replicated four outcomes from the Centers for Medicare and Medicaid Services Federally Qualified Health Center Advanced Primary Care Practice pilot. The DES model simulates a synthetic population's healthcare experience, including symptom onset, appointment scheduling, screening, and treatment, as well as the impact of physician training. A network of detailed event modules was developed from peer-reviewed literature. Synthetic patients' attributes modify the probability distr...
A model is developed for separation of multicomponent gas mixtures in a countercurrent hollow fib... more A model is developed for separation of multicomponent gas mixtures in a countercurrent hollow fiber membrane module. While the model's solution in countercurrent module usually involves in a time consuming iterative procedure, a proper initial guess is proposed for beginning the calculation and a simple procedure is introduced for correcting the guesses, hereby the CPU time is decreased essentially. The model's predictions are compared with the experimental data and a good agreement is achieved. By using data taken from a LAB unit in Isfahan, the proposed model is applied to investigate the feasibility of membrane process for hydrogen separation in this unit. It is revealed that a high hydrogen purity and recovery could be achieved in the permeate stream, while the required area increases at higher stage cut or higher permeate
Depending on the way of computer configuration and information access in the network, the network... more Depending on the way of computer configuration and information access in the network, the networks are divided into two main groups: Peer-to-Peer and Client-Server. In peer to peer networks, there is no specific server and no hierarchy in terms of computers. Authentication in P2P environment is owned by the public keys, and is done by establishing a secure mechanism and storage of cryptographic keys on a user device. The basic factor for password-based authentications in P2P systems includes registering a user account and making a connection. In this paper, an efficient scheme is presented to support logging in to systems based on users’ user name-password authentication and Client-Server systems. In addition, password authentication protocols to log in a P2P networks such as account registration, login, password change, re-login, log-out of a remote device, password retrieval, restarting the forgotten password via e-mail or security questions will be introduced.
Weakly Supervised Object Localization (WSOL) methods only require image level labels as opposed t... more Weakly Supervised Object Localization (WSOL) methods only require image level labels as opposed to expensive bounding box annotations required by fully supervised algorithms. We study the problem of learning localization model on target classes with weakly supervised image labels, helped by a fully annotated source dataset. Typically, a WSOL model is first trained to predict class generic objectness scores on an off-the-shelf fully supervised source dataset and then it is progressively adapted to learn the objects in the weakly supervised target dataset. In this work, we argue that learning only an objectness function is a weak form of knowledge transfer and propose to learn a classwise pairwise similarity function that directly compares two input proposals as well. The combined localization model and the estimated object annotations are jointly learned in an alternating optimization paradigm as is typically done in standard WSOL methods. In contrast to the existing work that learns...
Journal of Teaching Language Skills, 2015
This paper describes the development and validation of a new model and questionnaire to measure I... more This paper describes the development and validation of a new model and questionnaire to measure Iranian English as a foreign language learners’ attitudes towards the use of native versus non-native English language norms. Based on a comprehensive review of the related literature and interviews with domain experts, five factors were identified. A draft version of a questionnaire based on those five factors containing 40 items for assessing learners’ attitudes towards norms was designed. The draft version was piloted with a group of 273 Iranian learners and exploratory factor analysis (EFA) of the obtained data indicated that five factors could be extracted. Then the fitness of the model was checked through confirmatory factor analysis (CFA) through the administration of the questionnaire to another group of 554 Iranian English language learners. The result of CFA revealed that the model enjoyed a satisfactory level of fitness indices, meaning that the five-factor structure including...
Int. J. Netw. Secur., 2020
To establish secure channel for network communication in open and distributed environments, authe... more To establish secure channel for network communication in open and distributed environments, authenticated key agreement protocol is an important primitive for establishing session key. So far, a great deals of identity-based protocols have been proposed to provide secure mutual authentication and common session key establishment in two-party setting for secure communications in the open environment. Majority of the existing authenticated key agreement protocols only provide partial forward secrecy. Therefore, such protocols are unsuitable for real-world applications that require a stronger sense of perfect forward secrecy. In this paper, we present a secure twoparty identity-based authenticated key agreement protocol with achieves most of the required security attributes. We also show that the scheme achieves the security attributes include known-key secrecy, perfect forward secrecy, PKG forward secrecy, key-compromise impersonation resilience, unknown key-share resilience, no key r...
Uploads
Papers by Amir Hossein Rahimi