Ehsan Hosseini-Asl

email: <firstname>.<lastname> [at] gmail.com

Resume

Linkedin

Google scholar

Bio

I am a senior research scientist in Salesforce AI Research. My research area is deep learning with application to natural language processing, speech processing and computer vision. My current focus is building end-to-end language modeling in NLP, few-shot learning, and sample efficient deep learning models that generalize better to the distribution shift.

NEWS

• April, 2022: My paper on few-shot aspect-based sentiment analysis is accepted to NAACL 2022

• September, 2020: My paper on end-to-end language modeling for task-oriented dialogye was acceptet as spotlight in NeurIPS 2022

• October, 2019: Our on end-to-end belief state tracking for task-oriented dialogye was selected as outstanding paper at ACL 2019

• June, 2019: Our paper on end-to-end belief state tracking for task-oriented dialogye was accepter to ACL 2019

• December, 2018: My paper on few-shot domain adaptation was accepted to ICLR 2019

• June, 2018: My paper on speech synthesis using generative adversarial network is accepted to Interspeech 2018

Honors and Awards

• 2020 Spotlight paper at NeurIPS 2020
• 2019 Outstanding paper at ACL 2019
• 2016 Outstanding PhD Dissertation Award (John M. Houchens Prize), School of Interdisciplinary and Graduate Studies, University of Louisville
• 2016 Graduate Dean’s Citation School of Interdisciplinary and Graduate Studies, University of Louisville
• IEEE 2015 Graduate Student Research grant, IEEE Computational Intelligence Society, “Part-based Representation of Data In Deep Learning Models”

Experience

• Senior Research Scientist, Salesforce AI Research (March 2017-present)

Education

Ph.D. in Electrical and Computer Engineering, University of Louisville, USA (2012-2016)
Dissertation Title: “Sparse Feature Learning for Image Analysis in Segmentation, Classification and Disease Diagnosis”

• Advisor: Ayman El-Baz

Demo

Task-Oriented Dialogue

• SimpleTOD: An end-to-end Language model for Task-Orinted Dialogue (NeurIPS 2020 Spothlight) [code], [blog], [paper], [spotlight presentation]

Sentiment Analysis

• Few-shot Aspect-based Sentiment Analysis with generative language model (NAACL 2022) [code], [paper]

Adversarial Domain Adaptation / Speech

Augmented Cyclic Adversarial learning for low-resource domain adaptation, ICLR 2019, [paper]

GAN for Speech Syntesis

• Low-resource speech synthesis/recognition using Multi-Discriminator Generative Adversarial Network, Interspeech 2019 [paper], [blog]

Publication

Journals

• E. Hosseini-Asl, M. Ghazal, A. Mahmoud, A.M. Aslantas., M.F. Shalaby., G.N. Casanova, G. Barnes, R. G. Gimel’farb, R. Keynton and A. El-Baz, ”Alzheimer’s disease diagnostics by a 3D deeply supervised adaptable convolutional network.” Frontiers in bioscience (Landmark edition), 23, p.584., 2018

• E. Hosseini-Asl, J. M. Zurada, O. Nasraoui, “Deep Learning of Part-based Representation of Data Using Sparse Autoencoders with Nonnegativity Constraints”, Neural Networks and Learning Systems, IEEE Trans. on, vol.PP, no.99, pp.1-13, 2015. paper, code-matlab

• E. Hosseini-Asl, J. M. Zurada, Georgy Gimel’farb, and A. El-Baz, “3D Lung Segmentation Using Incremental Constrained Nonnegative Matrix Factorization,” Biomedical Engineering, IEEE Trans. on, vol.PP, no.99, pp.1-1, 2015. paper

Conference papers

Natural Language Processing

• E. Hosseini-Asl, W. Liu, C. Xiong, “A Generative Language Model for Few-shot Aspect-Based Sentiment Analysis”, NAACL 2022 (paper), (code)

• E. Hosseini-Asl, B. McCann, C. Wu, S. Yavuz, R. Socher, “A Simple Language Model for Task-Oriented Dialogue”, NeurIPS 2020 Spothlight, (paper), (code)

• T. He, B. McCann, C. Xiong, E. Hosseini-Asl, “Joint Energy-based Model Training for Better Calibrated Natural Language Understanding Models”, EACL 2021, (paper), (code)

• C. S. Wu, A. Madotto, E. Hosseini-Asl, C. Xiong, R. Socher, P. Fung, “Transferable multi-domain state generator for task-oriented dialogue systems”. ACL 2019 (outstanding paper), (paper), (code)

• E. Nouri, E. Hosseini-Asl, “Toward Scalable Neural Dialogue State Tracking Model”, NeurIPS 2018, 2nd Conversational AI workshop, (paper), (code)

• E. Hosseini-Asl, and A. Guha, “Similarity-based Text Recognition By Deeply-Supervised Siamese Network, 2015, arXiv:1511.04397 [cs.CV], (paper), (code-python/theano)

• E. Hosseini-Asl and J. M. Zurada, “Nonnegative Matrix Factorization for Document Clustering: A Survey,” in Artificial Intelligence and Soft Computing, Springer International Publishing, 2014, vol. 8468, pp. 726–737. paper

Speech processing

• E. Hosseini Asl, Y. Zhou, C. Xiong, R. Socher, “Augmented Cyclic Adversarial Learning for Low Resource Domain Adaptation “, ICLR 2019, (paper)

• E. Hosseini-Asl, Y.Zhou, C. Xiong, R. Socher, “A Multi-Discriminator CycleGAN for Unsupervised Non-Parallel Speech Domain Adaptation”, Interspeech 2018, (paper)

Medical Image Analysis

• E. Hosseini-Asl, G. Gimel’farb, and A. El-Baz, “Alzheimer’s Disease Diagnostics by a Deeply Supervised Adaptable 3D Convolutional Network”, arXiv:1607.00556 [cs.LG, q-bio.NC, stat.ML], (paper), (code-python/theano)

• E. Hosseini-Asl, R. Keynton, and A. El-Baz, “Alzheimer’s Disease Diagnosis by Adaptation of 3D Convolutional Network”, Image Processing (ICIP), 2016 IEEE Int. Conference on, Phoenix, Arizona, USA, September 25-28, 2016, (paper), (code-python/theano)

• M. Shehata, F. Khalifa, E. Hollis, A. Soliman, E. Hosseini-Asl, M. Abou El-Ghar, M. El-Baz, A. Dwyer, A. El-Baz, R. Keynton, “A New Non-Invasive Approach for Early Classification of Renal Rejection Types Using Diffusion-Weighted MRI”, Image Processing (ICIP), 2016 IEEE Int. Conference on, Phoenix, Arizona, USA, September 25-28, 2016.

• I. Reda, A. Shalaby, F. Khalifa, M. Elmogy, A. Aboulfotouh, M. Abou El-Ghar, E. Hosseini-Asl, N. Werghi, R. Keynton, A. El-Baz, “Computer-Aided Diagnosis Tool for Early Detection of Prostate Cancer”, Image Processing (ICIP), 2016 IEEE Int. Conference on, Phoenix, Arizona, USA, September 25-28, 2016.

• I. Reda, A. Shalaby, M. Abou El-Ghar, F. Khalifa, M. Elmogy, A. Aboulfotouh, E. Hosseini-Asl, A. El-Baz, and R. Keynton, “A New NMF-Autoencoder Based CAD System For Early Diagnosis of Prostate Cancer”, 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI), Prague, 2016, pp. 1237-1240. (paper)

• M. Ismail, M. Nitzken, A. E. Switala, E. Hosseini-Asl, M. Mahmoud, A. Shalaby, M. Casanova, A. El-Baz, “A New CAD System for Early Diagnosis of Autism Using Structural MRI”, ICIP, 2017. code-python/theano

• E. Hosseini-Asl, J. M. Zurada, A. El-baz, “Automatic Segmentation of Pathological Lung Using Incremental Nonnegative Matrix Factorization”, in Image Processing (ICIP), 2015 IEEE Int. Conference on, Quebec City, Canada, September 27-30, pp.3111-3115, 2015. paper

• E. Hosseini-Asl, J.M. Zurada, A. El-baz, “Lung Segmentation Based on Nonnegative Matrix Factorization,” in Image Processing (ICIP), 2014 IEEE Int. Conference on, Paris, France, Oct 2014, pp. 877-881. paper

Autoencoder & Sparse Feature Learning

• E. Hosseini-Asl, “Structured Sparse Convolutional Autoencoder”, arXiv:1604.04812 [cs.LG cs.NE](paper), (code-python/theano)

• B. Ayinde, E. Hosseini-Asl, J.M. Zurada, “Visualizing and Understanding Nonnegativity Constrained Sparse Autoencoder in Deep Learning.”, In International Conference on Artificial Intelligence and Soft Computing, 12 (pp. 3-14), June 2016. (paper)

• Zurada, Jacek M., Tolga Ensari, Ehsan Hosseini-Asl, Jan Chorowski, “Nonnegative Matrix Factorization and Its Application to Pattern Analysis and Text Mining.”, FEDCSIS 2013. paper

• E. Hosseini-Asl, J. M. Zurada, “Multiplicative Algorithm for Correntropy-Based Nonnegative Matrix Factorization”, Journal of Applied Computer Science Methods, 2014.

Wavelet Theory for Neural network

• E. Hossaini-asl, M. Shahbazian, “Nonlinear dynamic system control using wavelet neural network based on sampling theory”, IEEE International Conference on Systems, Man and Cybernetics, SMC 2009, Pages: 4502 – 4507, 2009

• E. Hossaini-asl, M. Shahbazian, K. Salahshoor, “Non uniform noisy data training using wavelet neural network based on sampling theory”, WSEAS Transactions on Systems, Volume 7, Issue 12, pp. 1381-1391, December 2008.

• E. Hossaini-asl, M. Shahbazian, K. Salahshoor, “Wavelet neural network based on sampling theory for non uniform noisy data”, Selected Papers from the WSEAS Conferences in Spain, pp.51-56, 2008

Patent

• Determining a Symbol Based on Visual Features of an Image, provisional application filed: 076409-8004.US00 [provisional]
• Patent Disclosure: A Novel Automated and Adaptable MRI-based Alzheimer’s Disease Diagnosis CAD System [non-provisional]