About Me

I am a Data Scientist working with the Conversational AI DS team at Uber. I received my PhD in Computer Science and Cognitive Science from Tufts University in Feb 2019.  Before coming to Tufts, I received a M.Sc. in Electrical and Computer Engineering from National University of Singapore and a B.Sc. in Computer Engineering from Iran University of Science and Technology. Broadly speaking, my research interests lie at the intersection of AI, machine learning and cognitive science. More specifically, I am interested in truly adaptive/data-driven NLP/AI systems.
My PhD research was informed by studying the developmental trajectory of intelligence in young infants who start with little or no knowledge of the world. In my PhD dissertation, I studied and developed local approaches to learning, (1) which depart from computationally expensive ideal learners, (2) without a huge loss in functional performance, (3) while replicating human performance, and (4) accounting for real-world constraints faced by learners. This dissertation consists of three main parts. The first part studies the problem of early word learning in isolation, without considering the interactions between word learning and the acquisition of other relevant capabilities. The second part of the dissertation studies the problem of word learning as the joint acquisition of word order and words’ referents, taking into account the interactions between word learning and the acquisition of syntax. We proposed that a selective focus on the words with concrete object or action referents and tracking the sequential patterns of the semantic roles associated with their referents can guide the acquisition of the notion of word order in the absence of any prior syntactic knowledge (e.g., knowledge of “subjecthood” or “objecthood”). The last part of the dissertation investigates the application of the pro- posed incremental and memory-limited models in realization of real-time word learning in robots. We added a new word learning component to the cognitive robotic architecture DIARC to enable the acquisition of new word-object associations in robots incrementally and in real-time. Additionally, we investigated the combination of cross-situational word learning and instruction-based word learning in DIARC which enables the robot to learn words from both human-human and human-robot interactions.


Sadeghi, S., Oosterveld, B., Krause, E., Scheutz, M. (accepted). Acquisition of Word-Object Associations from Human-Robot and Human-Human Dialogues. In Proceedings of the 2019 IEEE International Conference on Robotics and Automation (ICRA 2019). pdf (demo 1 and demo 2).

Sadeghi, S., Scheutz, M. (2018). Sensitivity to Input Order: Evaluation of an Incremental and Memory-Limited Bayesian Cross-Situational Word Learning Model. In Proceedings of the 27th International Conference on Computational Linguistics (COLING 2018).pdf data

Sadeghi, S., Scheutz, M. (2018). Early Syntactic Bootstrapping in an Incremental Memory-Limited Word Learner. In Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18). Awarded student travel scholarship. pdf

Sadeghi, S., Scheutz, M. (2017). Joint Acquisition of Word Order and Word Referent in a Memory-Limited and Incremental Learner. In Proceedings of the 2017 8th IEEE International Conference on Cognitive Infocommunications (CogInfoCom). pdf

Sadeghi, S., Scheutz, M., Krause, E. (2017). An Embodied Incremental Bayesian Model of Cross-Situational Word Learning. In Proceedings of the 2017 Joint IEEE International Conference on Development and learning and Epigenetic Robotics (ICDL-EPIROB). pdf (demo).

Scheutz, M., Krause, E., and Sadeghi, S. (2014). An Embodied Real-Time Model of Language-Guided Incremental Visual Search. In Proceedings of the 36th Annual Conference of the Cognitive Science Society. pdf

Sadeghi, S., Scheutz, M., Pu, H., Holcomb, P. J., and Midgley, K. J. (2013). A PDP Model for Capturing N400 Effects in Early L2 Learners during Bilingual Word Reading Tasks. In Proceedings of the 35th Annual Conference of the Cognitive Science Society. pdf

Sadeghi, S., and Ramanathan, K. (2011, July). A Hubel Wiesel model of early concept generalization based on local correlation of input features. In Proceedings of the 2011 IEEE International Joint Conference on Neural Networks. (pp. 709-716). pdf


Ramanathan, K and Sadeghi, S., Data structure and a method for using the data structure. US20130018832.

Unpublished Manuscripts

Sadeghi S., Incremental and Memory-Limited Probabilistic Generative Models of Early Word Learning. Ph.D. Thesis.

Sadeghi S., A Hubel Wiesel Model of Early Concept Generalization based on Local Correlation of Input Features. M.Eng. Thesis.

Sadeghi, S., Fusion of Color and Infrared images for pedestrian detection using SVM, B.Sc. Thesis.