Publications

Hassan Khosravi

The updated list of my publications can be found on my Google Scholar page

Reciprocal Content Recommendation for Peer Learning Study Sessions

Conference Paper 2018
Boyd Potts, Hassan Khosravi and Carl Reidsema (2018). Proceeding of the 19th International Conference on Artificial Intelligence in Education
Recognition of peer learning as a valuable supplement to formal education has lead to a rich literature formalising peer learning as an institutional resource. Facilitating peer learning support sessions alone however, without providing guidance or context, risks being ineffective in terms of any targeted, measurable effects on learning. Building on an existing open-source, student-facing platform called RiPPLE, which recommends peer study sessions based on the availability, competencies and compatibility of learners, this paper aims to supplement these study sessions by providing content from a repository of multiple-choice questions to facilitate topical discussion and aid productiveness. We exploit a knowledge tracing algorithm alongside a simple Gaussian scoring model to select questions that promote relevant learning and that reciprocally meet the expectations of both learners. Primary results using synthetic data indicate that the model works well at scale in terms of the number of sessions and number of items recommended, and capably recommends from a large repository the content that best approximates a proposed difficulty gradient.
@inproceedings{Potts:2018:RPR:3170358.3170400,
                  author = {Potts, Boyd A. and Khosravi, Hassan and Reidsema, Carl},
                  title = {Reciprocal Content Recommendation for Peer Learning Study Sessions},
                  booktitle = {Proceeding of the 19th International Conference on Artificial Intelligence in Education},
                  series = {AIED2018},
                  year = {2018},
                 } 
              


Reciprocal Peer Recommendation for Learning Purposes

Conference Paper 2018
Boyd Potts, Hassan Khosravi , Carl Reidsema, Aneesha Bakharia, Mark Belonogof, Melanie Fleming (2018). Proceeding of the 8th International Learning Analytics and Knowledge (LAK) Conference
Larger student intakes by universities and the rise of education through Massive Open Online Courses has led to less direct contact time with teaching staff for each student. One potential way of addressing this contact deficit is to invite learners to engage in peer learning and peer support; however, without technological support they may be unable to discover suitable peer connections that can enhance their learning experience. Two different research subfields with ties to recommender systems provide partial solutions to this problem. Reciprocal recommender systems provide sophisticated filtering techniques that enable users to connect with one another. To date, however, the main focus of reciprocal recommender systems has been on providing recommendation in online dating sites. Recommender systems for technology enhanced learning have employed and tailored exemplary recommenders towards use in education, with a focus on recommending learning content rather than other users. In this paper, we first discuss the importance of supporting peer learning and the role recommending reciprocal peers can play in educational settings. We then introduce our open-source course-level recommendation platform called RiPPLE that has the capacity to provide reciprocal peer recommendation. The proposed reciprocal peer recommender algorithm is evaluated against key criteria such as scalability, reciprocality, coverage, and quality and shows improvement over a baseline recommender. Primary results indicate that the system can help learners connect with peers based on their knowledge gaps and reciprocal preferences, with designed flexibility to address key limitations of existing algorithms identified in the literature.
@inproceedings{Potts:2018:RPR:3170358.3170400,
                                      author = {Potts, Boyd A. and Khosravi, Hassan and Reidsema, Carl and Bakharia, Aneesha and Belonogoff, Mark and Fleming, Melanie},
                                      title = {Reciprocal Peer Recommendation for Learning Purposes},
                                      booktitle = {Proceedings of the 8th International Conference on Learning Analytics and Knowledge},
                                      series = {LAK '18},
                                      year = {2018},
                                      isbn = {978-1-4503-6400-3},
                                      location = {Sydney, New South Wales, Australia},
                                      pages = {226--235},
                                      numpages = {10},
                                      url = {http://doi.acm.org/10.1145/3170358.3170400},
                                      doi = {10.1145/3170358.3170400},
                                      acmid = {3170400},
                                      publisher = {ACM},
                                      address = {New York, NY, USA},
                                      keywords = {peer learning, peer support, reciprocal recommender, recsysTEL},
                                     } 
                                     
                                  


Graph-based Visual Topic Dependency Models: Supporting Assessment Design and Delivery at Scale

Conference Paper 2018 Best Short Research Paper Nomination
Kendra Cooper and Hassan Khosravi (2018). Proceeding of the 8th International Learning Analytics and Knowledge (LAK) Conference
Educational environments continue to rapidly evolve to address the needs of diverse, growing student populations, while embracing advances in pedagogy and technology. In this changing landscape the design and delivery of high-quality assessments become even more challenging. In particular, ensuring that there is consistency among the assessments for di erent o erings of a course, providing meaningful feedback about students’ achievements, and tracking students’ progression over time are all challenging tasks at scale. Here, a collection of visual Topic Dependency Models (TDMs) is proposed that addresses the three facets of t he challenges noted above. It visualises the required topics and their dependencies at a course level (e.g., CS 100); visualises assessment achievement data at a classroom level (e.g., CS 100 Term 1 2016 Section 001) both at one point in time ( static) and over time (dynamic). The collection of TDMs share a common, two-weighted graph foundation; the models are illustrated using examples drawn from a high-enrolment introductory course on C programming for engineering students. An algorithm is presented to create a TDM (static achievement for a cohort). A proof of concept tool is under development; the current version is described brie y in terms of its support for visualising existing (historical, test) and synthetic data generated on demand.
 @inproceedings{Cooper:2018:GVT:3170358.3170418,
                                      author = {Cooper, Kendra and Khosravi, Hassan},
                                      title = {Graph-based Visual Topic Dependency Models: Supporting Assessment Design and Delivery at Scale},
                                      booktitle = {Proceedings of the 8th International Conference on Learning Analytics and Knowledge},
                                      series = {LAK '18},
                                      year = {2018},
                                      isbn = {978-1-4503-6400-3},
                                      location = {Sydney, New South Wales, Australia},
                                      pages = {11--15},
                                      numpages = {5},
                                      url = {http://doi.acm.org/10.1145/3170358.3170418},
                                      doi = {10.1145/3170358.3170418},
                                      acmid = {3170418},
                                      publisher = {ACM},
                                      address = {New York, NY, USA},
                                      keywords = {graph algorithms, student assessment, visual analytics},
                                     } 
                                  


Recommendation in Personalised Peer Learning Environments

Report 2017
Hassan Khosravi
Recommendation in Personalised Peer Learning Environments (RiPPLE) is an adaptive, crowdsourced, web-based, student-facing, open-source platform that employs exemplary techniques from the fields of machine learning, crowdsourcing, learning analytics and recommender systems to provide personalised content and learning support at scale. RiPPLE presents students with a repository of tagged multiple-choice questions and provides instant feedback in response to student answers. The repository of the questions is created in partnership with the students through the use of crowdsourcing. RiPPLE uses students responses to the questions to approximate their knowledge states. Based on their knowledge state and learning needs, each student is recommended a set of personalised questions. For students that are interested in providing learning support, seeking learning support or finding study partners, RiPPLE recommends peer learning sessions based on their availability, knowledge state and preferences. This paper describes the RiPPLE interface and an implementation of that interface that has been built at the University of Queensland. The RiPPLE platform and a reference implementation
 @article{khosravi2017recommendation,
                                  title={Recommendation in Personalised Peer-Learning Environments},
                                  author={Khosravi, Hassan},
                                  journal={arXiv preprint arXiv:1712.03077},
                                  year={2017}
                                }
                              


Analysing the Learning Pathways of Students in a Large Flipped Engineering Course

Conference Paper 2017
Carl Reidsema, Hassan Khosravi , Melanie Fleming, Nick Achilles, Esther Fink (2017). Proceedings of the 34th International Conference on Innovation, Practice and Research in the Use of Educational Technologies in Tertiary Education
Recent advancements in educational technologies (learning management systems, online discussion forums, peer-learning tools) coupled with new methods of course delivery (e.g. blended, flipped, MOOCs) provide significant opportunities for universities to deliver challenging, high quality, yet engaging curriculum for students. In this paper, we examine the variations and similarities of student’s approaches to learning (learning pathways) by examining how well they performed in a large (N ~ 1000 student) first year engineering flipped classroom. The analysis focused on student’s performance in their assessment (formative and summative) as well as their online interaction with a range of online tools purposely built to support students through peer learning and acquisition of resources and expertise. Analysis using k-means clustering reveals that students do in fact adopt a variety of successful pathways through the course. The unique aspects of this work lie in the use of analytics algorithms that whilst perhaps routinely utilised in data mining, are not as well utilised in better understanding patterns (successful or otherwise) of student interactions within a technology enhanced active learning environment that integrates theory with engineering practice.
 @inproceedings{Reidsema2017LearningPathways,
                                          title={Analysing the Learning Pathways of Students in a Large Flipped Engineering Course},
                                          author={Reidsema, Carl and Khosravi, Hassan and Fleming, Melanie and Achilles, Nick and Fink,  Esther},
                                          booktitle={Proceedings of the 34th International Conference of Innovation, Practice and Research in the Use of Educational Technologies in Tertiary Education},
                                          pages={372--382},
                                          year={2017},
                                        }


RiPLE: Recommendation in Peer-Learning Environments Based on Knowledge Gaps and Interests

Journal Paper 2017
Hassan Khosravi , Kendra Cooper, Kirsty Kitto (2017). Journal of Educational Data Mining
Various forms of Peer-Learning Environments are increasingly being used in post-secondary education, often to help build repositories of student generated learning objects. However, large classes can result in an extensive repository, which can make it more challenging for students to search for suitable objects that both reflect their interests and address their knowledge gaps. Recommender Systems for Technology Enhanced Learning (RecSysTEL) offer a potential solution to this problem by providing sophisticated filtering techniques to help students to find the resources that they need in a timely manner. Here, a new RecSysTEL for Recommendation in Peer-Learning Environments (RiPLE) is presented. The approach uses a collaborative filtering algorithm based upon matrix factorization to create personalized recommendations for individual students that address their interests and their current knowledge gaps. The approach is validated using both synthetic and real data sets. The results are promising, indicating RiPLE is able to provide sensible personalized recommendations for both regular and cold-start users under reasonable assumptions about parameters and user behavior.
 @article{khosravi2017,
                                    title={RiPLE: Recommendation in Peer-Learning Environments Based on Knowledge Gaps and Interests},
                                    author={Khosravi, Hassan and Cooper, Kendra and Kitto, Kirsty},
                                    journal={JEDM-Journal of Educational Data Mining},
                                    volume={9},
                                    number={1},
                                    pages={42-67},
                                    year={2017}
                                  }
                                  


Using Learning Analytics to Investigate Patterns of Performance and Engagement in Large Classes

Conference Paper 2017
Hassan Khosravi , Kendra Cooper (2017). Proceedings of the Special Interest Group on Computer Science Education conference.
Educators continue to face significant challenges in providing high quality, post-secondary instruction in large classes including: motivating and engaging diverse populations (e.g., academic ability and backgrounds, generational expectations); and providing helpful feedback and guidance. Researchers investigate solutions to these kinds of challenges from alternative perspectives, including learning analytics (LA). Here, LA techniques are applied to explore the data collected for a large, flipped introductory programming class to (1) identify groups of students with similar patterns of performance and engagement; and (2) provide them with more meaningful appraisals that are tailored to help them effectively master the learning objectives. Two studies are reported, which apply clustering to analyze the class population, followed by an analysis of a subpopulation with extreme behaviours.
 @inproceedings{khosravi2017largeclasses,
                                          title={Using Learning Analytics to Investigate Patterns of Performance and Engagement in Large Classes},
                                          author={Khosravi, Hassan and Cooper, Kendra ML},
                                          booktitle={Proceedings of the 2017 ACM SIGCSE Technical Symposium on Computer Science Education},
                                          pages={309--314},
                                          year={2017},
                                          organization={ACM}
                                        }


An Overview of Learning Analytics in Higher Education

Report 2017
Amy Wong, Marcel Lavrencic, Hassan Khosravi (2017) Institute for Teaching and Learning Innovation, The University of Queensland
This paper provides an overview of the recent national and international initiatives, adoptions, and challenges around learning analytics. It aims to assist the Learning Analytics Steering Group at the University of Queensland in setting the context and priorities in the adoption of learning analytics, as we embrace the opportunities and challenges big data creates in higher education.


Enhanced security authentication methods, systems and media

Patent 2016
Andrew Csinger, Ildar Muslukhov, Hassan Khosravi, Peter Luong (2016)
A transaction authorization apparatus includes a processor in communication with a communications interface. The processor is configured to receive a request for a transaction requested by a user with whom a plurality of user devices are associated, to obtain respective transaction measurements from at least some available devices from among the plurality of user devices, and to confirm approval of the request for the transaction in response to confirmation that the transaction measurements satisfy a multi­device authorization policy associated with the transaction.
 @misc{csinger2016enhanced,
                                        title={Enhanced security authentication methods, systems and media},
                                        author={Csinger, A. and MUSLUKHOV, I. and KHOSRAVI, H. and LUONG, P.T.},
                                        url={http://www.google.com.pg/patents/WO2016115633A1?cl=en},
                                        year={2016},
                                        month=jul # "~28",
                                        publisher={Google Patents},
                                      }

                                    


Retiring the Red Pen: Marking Exams Digitally

Poster 2016
Hassan Khosravi, Donald Acton (2016), UBC Annual Science Education Open House
Over the last several years members of the Computer Science department have been marking exams, scanning them, recording grades using optical character recognition, and then returning online digital copies to students. This has greatly decreased the amount of time spent handing back exams and recording marks. With the advent of cloud-based software like Gradescope and Crowdmark two of us decided to switch the workflow and load scanned exams into grading software and then marked them online. During the 2015W2 term we collectively marked over 1500 exams (including group exams), across 3 courses (APSC 160, CPSC 313, and CPSC 317) utilizing over 30 TAs.


Method and System for Decoupling User Authentication and Data Encryption on Mobile Devices

Patent 2015
Hassan Khosravi, Ildar Muslukhov, Peter Luong (2015)
Over a billion people today use smartphones - a portable and highly mobile personal computer. Various data is stored on such devices for the benefit of being accessible "on-the-go". This creates the necessity to protect sensitive data that are stored on smartphones, i.e., "data-at-rest". Data encryption with a randomly generated encryption key can be applied. However, since the device has to be able to work in off-line scenarios, the encryption key has to be stored on the device, along with data. In order to overcome this limitation, all major mobile platforms encrypt the encryption key with so called key encryption key. The key encryption key is derived from a secret that is used to authenticate smartphone users (e.g., a PIN-code, or password). Unfortunately, PIN-codes are weak and are prone to brute-force attacks. In this work we present the KeyVault system, that decouples user authentication and data encryption. The results of the evaluation suggest that the KeyVault system is a practical solution that requires low maintainability from users.
@misc{khosravi2015coupling,
                                        title={Method and system for decoupling user authentication and data encryption on mobile devices},
                                        author={Khosravi, Hassan and Muslukhov, Ildar and Luong, Peter},
                                        year={2015},
                                        month=sep # "~15",
                                        publisher={Google Patents},
                                        note={US Patent 9,137,659}
                                      }

                                    


Modelling Relational Statistics With Bayes Nets Using Pseudo-Likelihood

Journal Paper 2014
Oliver Schulte, Hassan Khosravi , Arthur Kirkpatrick, Tong Man, Tianxiang Gao, Yi Xiong, Yuke Zhu, Jianxing Ling (2014). Journal of Machine Learning.
Learning statistical patterns based on relational frequencies is a machine learning task that supports applications like strategic planning and query optimization. Bayes nets are a widely used generative model class for representing statistical patterns. Bayes net parameters are often learned using classic maximum likelihood estimation. Applying this to relational data is difficult, however, due to the dependencies introduced by cyclic relationships. To address cyclic relations, a Bayes Net pseudolikelihood measure was recently introduced for relational data. This measure can be used to construct maximum pseudolikelihood estimation for estimating Bayes Net parameters. A naive implementation is intractable due to the complexity imposed by negated relational links. We render this computation tractable by using the fast Mobius transform. Evaluation on four benchmark datasets shows that MPLE is consistent, and provides accurate
 @article{Schulte2014statisticalstats,
                                        author = {Schulte, Oliver and Khosravi, Hassan and Kirkpatrick, Arthur E. and Gao, Tianxiang and Zhu, Yuke},
                                        title = {Modelling Relational Statistics with Bayes Nets},
                                        journal = {Mach. Learn.},
                                        issue_date = {January   2014},
                                        volume = {94},
                                        number = {1},
                                        month = jan,
                                        year = {2014},
                                        issn = {0885-6125},
                                        pages = {105--125},
                                        numpages = {21},
                                        url = {http://dx.doi.org/10.1007/s10994-013-5362-7},
                                        doi = {10.1007/s10994-013-5362-7},
                                        acmid = {2558593},
                                        publisher = {Kluwer Academic Publishers},
                                        address = {Hingham, MA, USA},
                                        keywords = {Bayes Nets, M\"{o}bius transform, Pseudo-likelihood, Statistical-relational learning, Structured data},
                                        } 
                                    


Transaction-Based Link Strength Prediction Using Matrix Factorization

Conference Paper 2013
Hassan Khosravi , Ali Bozorgkhan, Oliver Schulte (2013). Computational Intelligence and Data Mining (CIDM), IEEE.
The revolution of online social networks and methods of analyzing them have attracted interest in many research fields. Predicting whether a friendship holds in a social network between two individuals or not, link prediction, has been a heavily researched topic in the last decade. In this paper we research a related problem, link strength prediction, which aims to assign ratings or strengths to friendship links. A basic approach would be matrix factorization applied to friendship ratings. However, the existence of extensive transactions among users may be used for better predictions. We propose a new type of multiple-matrix factorization model for incorporating a transaction matrix. We derive gradient descent update equations for learning latent factors that predict values in the target rating matrix. To evaluate the model, we introduce data from Cloob which is a popular Iranian social network as well as synthetic data.
 @inproceedings{khosravi2013transaction,
                                          title={Transaction-based link strength prediction in a social network},
                                          author={Khosravi, Hassan and Bozorgkhan, Ali and Schulte, Oliver},
                                          booktitle={Computational Intelligence and Data Mining (CIDM), 2013 IEEE Symposium on},
                                          pages={191--198},
                                          year={2013},
                                          organization={IEEE}
                                        }
                                    


Fast Learning of Markov Logic Networks Via Moralization

Conference Paper 2012
Hassan Khosravi , Oliver Schulte, Tong Man (2012). proceeding of Inductive Logic programming .
Scaling MLN learning to medium-sized datasets with many descriptive attributes is a challenge. For structure learning, one of the most efficient and effective methods is the moralization approach, which uses Bayes net algorithms to learn a directed graphical model, then converts to an undirected one. Previously the MLN parameters were then learned using current MLN methods. In this paper we extend the moralization approach to parameter learning in addition to structure learning, where MLN weights are directly inferred from Bayes net CP-table entries. Empirical evaluation indicates that this approach to parameter learning is orders of magnitude faster while performing as well or better on prediction metrics.
 @inproceedings{khosravi2012fastlearning,
                                        title={Fast parameter learning for markov logic networks using bayes nets},
                                        author={Khosravi, Hassan},
                                        booktitle={International Conference on Inductive Logic Programming},
                                        pages={102--115},
                                        year={2012},
                                        organization={Springer}
                                      }
                                    


Learning Graphical Models for Relational Data via Lattice Search

Journal Paper 2012
Oliver Schulte, Hassan Khosravi (2012). Journal of Machine Learning .
This paper is an extension on "Structure Learning for Markov Logic Networks with Many Descriptive Attributes". In this paper we introduce a lattice search framework to present an efficient new algorithm for learning a Parametrized Bayes Net that performs a level-wise search through the table joins for relational dependencies. The previous version did not use the lattice search framework. Our new version adds constraints on the model search, makes all constraints explicit, and provides rationales and discussion of each. We have also added more comparison with other Markov Logic Network learning methods (e.g., BUSL, LSM), with well known inductive logic programming methods and a lesion study that assesses the effects of using only part of the components of our main algorithm.
 @article{schulte2012learning,
                                        title={Learning graphical models for relational data via lattice search},
                                        author={Schulte, Oliver and Khosravi, Hassan},
                                        journal={Machine Learning},
                                        volume={88},
                                        number={3},
                                        pages={331--368},
                                        year={2012},
                                        publisher={Springer}
}
                                    


Learning Compact Markov Logic Networks With Decision Trees

Journal Paper 2012
Hassan Khosravi, Oliver Schulte, Jianfeng Hu, and Tianxiang Gao (2012) Journal of Machine Learning .
Statistical-relational learning combines logical syntax with probabilistic methods. Markov Logic Networks (MLNs) are a prominent model class that generalizes both first-order logic and undirected graphical models (Markov networks). The qualitative component of an MLN is a set of clauses and the quantitative component is a set of clause weights. Generative MLNs model the joint distribution of relationships and attributes. A state-of-the-art structure learning method is the moralization approach: learn a set of directed Horn clauses, then convert them to conjunctions to obtain MLN clauses. The directed clauses are learned using Bayes net methods. The moralization approach takes advantage of the high-quality inference algorithms for MLNs and their ability to handle cyclic dependencies. A weakness of the moralization approach is that it leads to an unnecessarily large number of clauses. In this paper we show that using decision trees to represent conditional probabilities in the Bayes net is an effective remedy that leads to much more compact MLN structures. In experiments on benchmark datasets, the decision trees reduce the number of clauses in the moralized MLN by a factor of 5-25, depending on the dataset. The accuracy of predictions is competitive with the unpruned model and in many cases superior.
 @article{khosravi2012learningcompact,
                                        title={Learning compact Markov logic networks with decision trees},
                                        author={Khosravi, Hassan and Schulte, Oliver and Hu, Jianfeng and Gao, Tianxiang},
                                        journal={Machine Learning},
                                        volume={89},
                                        number={3},
                                        pages={257--277},
                                        year={2012},
                                        publisher={Springer}
                                        }
                                    


Learning Directed Relational Models With Recursive Dependencies

Journal Paper 2012
Oliver Schulte, Hassan Khosravi, Tong Man(2012) Journal of Machine Learning .
Recently, there has been an increasing interest in generative models that represent probabilistic patterns over both links and attributes. A key characteristic of relational data is that the value of a predicate often depends on values of the same predicate for related entities. For directed graphical models, such recursive dependencies lead to cycles, which violates the acyclicity constraint of Bayes nets. In this paper we present a new approach to learning directed relational models which utilizes two key concepts: a pseudo likelihood measure that is well defined for recursive dependencies, and the notion of stratification from logic programming. An issue for modelling recursive dependencies with Bayes nets are redundant edges that increase the complexity of learning. We propose a new normal form format that removes the redundancy, and prove that assuming stratification, the normal form constraints involve no loss of modelling power. Empirical evaluation compares our approach to learning recursive dependencies with undirected models (Markov Logic Networks). The Bayes net approach is orders of magnitude faster, and learns more recursive dependencies, which lead to more accurate predictions. A preliminary version of this work was published in the proceeding of Inductive Logic programming 2011
 @article{schulte2012learningrecursive,
                                          title={Learning directed relational models with recursive dependencies},
                                          author={Schulte, Oliver and Khosravi, Hassan and Man, Tong},
                                          journal={Machine learning},
                                          volume={89},
                                          number={3},
                                          pages={299--316},
                                          year={2012},
                                          publisher={Springer}
                                        }
                                    


Directed Models For Statistical Relational Learning

Thesis 2012
Hassan Khosravi
Statistical Relational Learning is a new branch of machine learning that aims to model a joint distribution over relational data. Relational data consists of different types of objects where each object is characterized with a different set of attributes. The structure of relational data presents an opportunity for objects to carry additional information via their links and enables the model to show correlations among objects and their relationships. This dissertation focuses on learning graphical models for such data. Learning graphical models for relational data is much more challenging than learning graphical models for propositional data. One of the challenges of learning graphical models for relational data is that relational data, unlike propositional data, is non independent and identically distributed and cannot be viewed in a single table. Relational data can be modeled using a graph, where objects are the nodes and relationships between the objects are the edges. In this graph, there may be multiple edges between two nodes because objects may have different types of relationships with each other. The existence of multiple paths of different length among objects makes the learning procedure much harder than learning from a single table. We use a lattice search approach with lifted learning to deal with the multiple path problem. We focus on learning the structure of Markov Logic Networks, which are a first order extension of Markov Random Fields. Markov Logic Networks are a prominent undirected statical relational model that have achieved impressive performance on a variety of statistical relational learning tasks. Our approach combines the scalability and efficiency of learning in directed relational models, and the inference power and theoretical foundations of undirected relational models. We utilize an extension of Bayesian networks based on first order logic for learning class-level or first-order dependencies, which model the general database statistics over attributes of linked objects and their links. We then convert this model to a Markov Logic Network using the standard moralization procedure. Experimental results indicate that our methods are two orders of magnitude faster than, and predictive metrics are superior or competitive with, state-of-the-art Markov Logic Network learners.
 @phdthesis{khosravi2012phdthesis,
                                        title={Directed Models For Statistical Relational Learning},
                                        author={Khosravi, Hassan},
                                        year={2012},
                                        school={Simon Fraser University}
                                      }
                                    


Structure Learning for Markov Logic Networks with Many Descriptive Attributes

Conference Paper 2010
Hassan Khosravi Oliver Schulte, Tong Man, Xiaoyuan Xu, and Bahareh Bina (2010). Proceedings of the Twenty-Fourth Conference on Artificial Intelligence
Many machine learning applications that involve relational databases incorporate first-order logic and probability. Many of the current state-of-the-art algorithms for learning MLNs have focused on relatively small datasets with few descriptive attributes, where predicates are mostly binary and the main task is usually prediction of links between entities. This paper addresses what is in a sense a complementary problem: learning the structure of a graphical model that models the distribution of discrete descriptive attributes given the links between entities in a relational database. Descriptive attributes are usually non-binary and can be very informative, but they increase the search space of possible candidate clauses. We present an efficient new algorithm for learning a directed relational model (parametrized Bayes net). From the Bayes net we obtain an MLN structure via a standard moralization procedure for converting directed models to undirected models. Learning MLN structure by moralization is 200-1000 times faster and scores substantially higher in predictive accuracy than benchmark MLN algorithms on three relational databases.
 @inproceedings{Khosravi2010structure,
                                        author = {Khosravi, Hassan and Schulte, Oliver and Man, Tong and Xu, Xiaoyuan and Bina, Bahareh},
                                        title = {Structure Learning for Markov Logic Networks with Many Descriptive Attributes},
                                        booktitle = {Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence},
                                        series = {AAAI'10},
                                        year = {2010},
                                        location = {Atlanta, Georgia},
                                        pages = {487--493},
                                        numpages = {7},
                                        url = {http://dl.acm.org/citation.cfm?id=2898607.2898686},
                                        acmid = {2898686},
                                        publisher = {AAAI Press},
                                        } 
                                    


A Survey on Statistical Relational Learning

Conference Paper 2010
Hassan Khosravi, and Bahareh Bina (2010) In Proceedings of Canadian Conference of AI
The vast majority of work in Machine Learning has focused on propositional data which is assumed to be identically and independently distributed, however, many real world datasets are relational and most real world applications are characterized by the presence of uncertainty and complex relational structure where the data distribution is neither identical nor independent. An emerging research area, Statistical Relational Learning (SRL), attempts to represent, model, and learn in relational domain. Currently, SRL is still at a primitive stage in Canada, which motivates us to conduct this survey as an attempt to raise more attention to this ?eld. Our survey presents a brief introduction to SRL and a comparison with conventional learning approaches. In this survey we review four SRL models as an attempt to raise more attention to this field. Our survey presents a brief introduction to SRL and a comparison with conventional learning approaches. In this survey we review four SRL models (PRMs, MLNs, RDNs, and BLPs) and compare them theoretically with respect to their representation, structure learning, parameter learning, and inference methods. We conclude with a discussion on limitations of current methods
 @inproceedings{khosravi2010survey,
                                        title={A Survey on Statistical Relational Learning},
                                        author={Khosravi, Hassan and Bina, Bahareh},
                                        booktitle={Advances in Artificial Intelligence: 23rd Canadian Conference on Artificial Intelligence, Canadian AI 2010, Ottawa, Canada, May 31-June 2, 2010, Proceedings},
                                        volume={6085},
                                        pages={256},
                                        year={2010},
                                        organization={Springer}
                                      }
                                    


The Imap Hybrid Method for Learning Gaussian Bayes Nets

Conference Paper 2010 Best paper award
Oliver Schulte, Gustavo Frigo, Russell Greiner, Hassan Khosravi (2010) In Proceedings of Canadian Conference of AI
This paper presents the I-map hybrid algorithm for selecting, given a data sample, a linear Gaussian model whose structure is a directed graph. The algorithm performs a local search for a model that meets the following criteria: (1) The Markov blankets in the model should be consistent with dependency information from statistical tests. (2) Minimize the number of edges subject to the ?rst constraint. (3) Maximize a given score function subject to the ?rst two constraints. Our local search is based on Graph Equivalence Search (GES); we also apply the recently developed SIN statistical testing strategy to help avoid local minima. Simulation studies with GES search and the BIC score provide evidence that for nets with 10 or more variables, the hybrid method selects simpler graphs whose structure is closer to the target graph.
 @inproceedings{schulte2010imap,
                                        title={The IMAP hybrid method for learning Gaussian Bayes nets},
                                        author={Schulte, Oliver and Frigo, Gustavo and Greiner, Russell and Khosravi, Hassan},
                                        booktitle={Canadian Conference on Artificial Intelligence},
                                        pages={123--134},
                                        year={2010},
                                        organization={Springer}
                                      }
                                    


A social network model of investment behaviour in the stock market

Journal Paper 2010
L Bakker, W Hare, Hassan Khosravi, B Ramadanovic (2010) Physica A: Statistical Mechanics and its Applications
This paper presents the I-map hybrid algorithm for selecting, given a data sample, a linear Gaussian model whose structure is a directed graph. The algorithm performs a local search for a model that meets the following criteria: (1) The Markov blankets in the model should be consistent with dependency information from statistical tests. (2) Minimize the number of edges subject to the ?rst constraint. (3) Maximize a given score function subject to the ?rst two constraints. Our local search is based on Graph Equivalence Search (GES); we also apply the recently developed SIN statistical testing strategy to help avoid local minima. Simulation studies with GES search and the BIC score provide evidence that for nets with 10 or more variables, the hybrid method selects simpler graphs whose structure is closer to the target graph.
 @article{bakker2010social,
                                        title={A social network model of investment behaviour in the stock market},
                                        author={Bakker, Laurens and Hare, Warren and Khosravi, Hassan and Ramadanovic, Bojan},
                                        journal={Physica A: Statistical Mechanics and its Applications},
                                        volume={389},
                                        number={6},
                                        pages={1223--1229},
                                        year={2010},
                                        publisher={Elsevier}
                                      }
                                    


Class-Level Bayes Nets for Relational Data

Report 2009
Oliver Schulte, Hassan Khosravi, Flavia Moser, Martin Ester
This paper presents the I-map hybrid algorithm for selecting, given a data sample, a linear Gaussian model whose structure is a directed graph. The algorithm performs a local search for a model that meets the following criteria: (1) The Markov blankets in the model should be consistent with dependency information from statistical tests. (2) Minimize the number of edges subject to the ?rst constraint. (3) Maximize a given score function subject to the ?rst two constraints. Our local search is based on Graph Equivalence Search (GES); we also apply the recently developed SIN statistical testing strategy to help avoid local minima. Simulation studies with GES search and the BIC score provide evidence that for nets with 10 or more variables, the hybrid method selects simpler graphs whose structure is closer to the target graph.
 @article{schulte2009class,
                                          author    = {Oliver Schulte and
                                                      Hassan Khosravi and
                                                      Flavia Moser and
                                                      Martin Ester},
                                          title     = {Join Bayes Nets: {A} new type of Bayes net for relational data},
                                          journal   = {CoRR},
                                          volume    = {abs/0811.4458},
                                          year      = {2008},
                                          url       = {http://arxiv.org/abs/0811.4458},
                                          timestamp = {Wed, 07 Jun 2017 14:40:54 +0200},
                                          biburl    = {http://dblp.uni-trier.de/rec/bib/journals/corr/abs-0811-4458},
                                          bibsource = {dblp computer science bibliography, http://dblp.org}
                                        }
                                    


LNBC: A Link-Based Naive Bayes Classifier

Conference Paper 2009
Bahareh Bina, Oliver Schulte, Hassan Khosravi (2009) IEEE International Conference on Data Mining Workshops
Many databases store data in relational format, with different types of entities and information about links between the entities. Link-based classification is the problem of predicting the class label of a target entity given information about features of the entity and about features of the related entities. A natural approach to link-based classification is to upgrade standard classification methods from the propositional, single-table testing. In this paper we propose a new classification rule for upgrading naive Bayes classifiers (NBC). Previous work on relational NBC has achieved the best results with link independency assumption which says that the probability of each link to an object is independent from the other links to the object. We formalize our method by breaking it into two parts: (1) the independent influence assumption: that the influence of one path from the target object to a related entity is independent of another. We consider object-path independency and (2) the independent feature assumption of NBC: that features of the target entity and a related entity are probabilistically independent given a target class label. We derive a new relational NBC rule that places more weight on the target entity features than formulations of the link independency assumption. The new NBC rule yields higher accuracies on three benchmark datasets-Mutagenesis, MovieLens, and Cora-with average improvements ranging from 2% to 10%
 @inproceedings{bina2009lnbc,
                                        title={LNBC: A Link-Based Naive Bayes Classifier},
                                        author={Bina, Bahareh and Schulte, Oliver and Khosravi, Hassan},
                                        booktitle={Data Mining Workshops, 2009. ICDMW'09. IEEE International Conference on},
                                        pages={489--494},
                                        year={2009},
                                        organization={IEEE}
                                      }
                                                                          


A New Hybrid Method for Bayesian Network Learning With Dependency Constraints

Conference Paper 2009
Oliver Schulte, Gustavo Frigo, Russell Greiner, Wei Luo, Hassan Khosravi (2009). Proceedings IEEE CIDM Symposium
A Bayes net has qualitative and quantitative aspects: The qualitative aspect is its graphical structure that corresponds to correlations among the variables in the Bayes net. The quantitative aspects are the net parameters. This paper develops a hybrid criterion for learning Bayes net structures that is based on both aspects. We combine model selection criteria measuring data fit with correlation information from statistical tests. We rely on the statistical test only to accept conditional dependencies, not conditional independencies. We show how to adapt local search algorithms to accommodate the observed dependencies. Simulation studies with GES search and the BDeu/BIC scores provide evidence that the additional dependency information leads to Bayes nets that better fit the target model in distribution and structure.
 @inproceedings{schulte2009hybrid,
                                          title={A new hybrid method for Bayesian network learning with dependency constraints},
                                          author={Schulte, Oliver and Frigo, Gustavo and Greiner, Russell and Luo, Wei and Khosravi, Hassan},
                                          booktitle={Computational Intelligence and Data Mining, 2009. CIDM'09. IEEE Symposium on},
                                          pages={53--60},
                                          year={2009},
                                          organization={IEEE}
                                        }
                                  


Exploratory Analysis of Co-Change Graphs for Code Refactoring

Conference Paper 2009
Hassan Khosravi and Recep Colak (2008). In Proceedings of Canadian conference of AI
A Bayes net has qualitative and quantitative aspects: The qualitative aspect is its graphical structure that corresponds to correlations among the variables in the Bayes net. The quantitative aspects are the net parameters. This paper develops a hybrid criterion for learning Bayes net structures that is based on both aspects. We combine model selection criteria measuring data fit with correlation information from statistical tests. We rely on the statistical test only to accept conditional dependencies, not conditional independencies. We show how to adapt local search algorithms to accommodate the observed dependencies. Simulation studies with GES search and the BDeu/BIC scores provide evidence that the additional dependency information leads to Bayes nets that better fit the target model in distribution and structure.
 @inproceedings{khosravi2009exploratory,
                                        title={Exploratory Analysis of Co-Change Graphs for Code Refactoring},
                                        author={Khosravi, Hassan and Colak, Recep},
                                        booktitle={Advances in Artificial Intelligence: 22nd Canadian Conference on Artificial Intelligence, Canadian AI 2009, Kelowna, Canada, May 25-27, 2009 Proceedings},
                                        volume={5549},
                                        pages={219},
                                        year={2009},
                                        organization={Springer}
                                      }
                                  


TACtic- A Multi Behavioral Agent for Trading Agent Competition

Conference Paper 2008
Hassan Khosravi M. E.shiri, Hamid. Khosravi, E. Iranmanesh A. Davoodi (2008). In Advances in Computer Science and Engineering, Communications in Computer and Information Science
Software agents are increasingly being used to represent humans in online auctions. Such agents have the advantages of being able to systematically monitor a wide variety of auctions and then make rapid decisions about what bids to place in what auctions. They can do this continuously and repetitively without losing concentration. To provide a means of evaluating and comparing (benchmarking) research methods in this area the trading agent competition (TAC) was established. This paper describes the design, of TACtic. Our agent uses multi behavioral techniques at the heart of its decision making to make bidding decisions in the face of uncertainty, to make predictions about the likely outcomes of auctions, and to alter the agent's bidding strategy in response to the prevailing market conditions.
 @incollection{khosravi2008tactic,
                                          title={TACtic-A Multi Behavioral Agent for Trading Agent Competition},
                                          author={Khosravi, Hassan and Shiri, Mohammad E and Khosravi, Hamid and Iranmanesh, Ehsan and Davoodi, Alireza},
                                          booktitle={Advances in Computer Science and Engineering},
                                          pages={811--815},
                                          year={2008},
                                          publisher={Springer}
                                        }
                                  


An Automated Negotiation Technique for Self-interest Agents

Conference Paper 2006
Hassan Khosravi, Mohammad Shiri, Hamid Khosravi, Ehsan Iranmanesh (2006) 12th International CSI Computer Conference
Due to the rapid growth of electronic environments, much research is currently being performed on autonomous trading mechanisms. Also in the last few years there has been an increasing interest from the agent community in the use of techniques from decision theory and game theory. Our paper connects theses two fields. In real world, when two agents negotiate, it seems rational that they emphasize on their highest priorities first. We give a strategyin which an agent can gain more profit by sacrificing or delaying some high priority tasks. The focus of this paper is on negotiation of self interest agents.