Barkan, O., Shaked, T., Fuchs, Y., & Koenigstein, N. (2023). Modeling Users' Heterogeneous Taste with Diversified Attentive User Profiles. Accepted for publication in User Modeling and User-Adapted Interaction.
Gaiger, K., Barkan, O., Tsipory-Samuel, S., & Koenigstein, N. (2023). Not All Memories Created Equal: Dynamic User Representations for Collaborative Filtering. IEEE Access, 11, 34746-34763. Gaiger, K., Barkan, O., Tsipory-Samuel, S., & Koenigstein, N. (2023). IEEE Access, 11, 34746-34763.
Katz, O., Barkan, O., Koenigstein, N., & Zabari, N. (2022, September). Learning to Ride a Buy-Cycle: A Hyper-Convolutional Model for Next Basket Repurchase Recommendation. In Proceedings of the 16th ACM Conference on Recommender Systems (pp. 316-326).
Eytan, L., Bogina, V., Ben-Gal, I., & Koenigstein, N. Mind the Path: Path-Based Knowledge-Graph with Neural Attention. Available at SSRN 4429484.
Barkan, O., Fuchs, Y., Caciularu, A., & Koenigstein, N. (2020, September). Explainable recommendations via attentive multi-persona collaborative filtering. In Proceedings of the 14th ACM Conference on Recommender Systems (pp. 468-473).
Explainable AI (XAI)
Oren Barkan, Yehonatan Elisha, Yuval Asher, Amit Eshel, Noam Koenigstein. “Visual Explanations via Iterated Integrated Attributions”. International Conference on Computer Vision (ICCV), October 2023, Paris, France.
Oren Barkan, Yehonatan Elisha, Jonath Weill, Yuval Asher, Amit Eshel, Noam Koenigstein. “Deep Integrated Explanations”. ACM International Conference on Information and Knowledge Management (CIKM), Birmingham, UK, November, 2023.
Oren Barkan, Yuval Asher, Amit Eshel, Yehonatan Elisha, Noam Koenigstein. “Learning To Explain: A Model-Agnostic Framework for Explaining Black Box Models”. IEEE International Conference on Data Mining (ICDM), Shanghai, China, December 2023.
Oren Barkan, Yehonatan Elisha, Jonathan Weill, Yuval Asher, Amit Eshel, Noam Koenigstein. “Stochastic Integrated Explanations for Vision Models”. IEEE International Conference on Data Mining (ICDM), Shanghai, China, December 2023.
Barkan, O., Benchimol, J., Caspi, I., Cohen, E., Hammer, A., & Koenigstein, N. Forecasting CPI Inflation Components with Hierarchical Recurrent Neural Networks. International Journal of Forecasting.
Machine Learning for Agricultural Improvements
Abiola Owoyemi, Ron Porat, Amnon Lichter, Adi Doron-Faigenboim, Tamar Holder, Noam Koenigstein, Yael Salzer. “Temperature interruptions harm the quality of stored 'Rustenburg' navel oranges and development of dynamic shelf-life prediction models”. Accepted for publication in Postharvest Biology and Technology.
Machine Learning for Electronic Music Production
Barkan, O., Tsiris, D., Katz, O., & Koenigstein, N. (2019). Inversynth: Deep estimation of synthesizer parameter configurations from audio signals. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 27(12), 2385-2396.
DELTA Lab - The lab that makes a difference!
The Deep Learning Technologies and Applications (DELTA) is an academic lab in the Department of Industrial Engineering at Tel Aviv University. DELTA Lab focuses on a broad spectrum of practical applications for deep learning algorithms that address real-world challenges
Department of Industrial Engineering - Tel Aviv University
The Deep Learning Technologies and Applications (DELTA) lab concentrates on a broad spectrum of practical applications for deep learning algorithms that address real-world challenges. Our research encompasses a diverse range of areas including, but not limited to, recommender systems, Explainable AI (XAI), machine learning for agricultural improvements, machine learning for macro-economics, integrating machine learning within education, and employing machine learning techniques in the production of electronic music. In our pursuit of advancing applied machine learning research, we establish robust partnerships with entities in both the industrial and public sectors.
Research on recommender systems forms a core part of our investigative pursuits at DELTA Lab. We delve into an array of study areas within this discipline that encompasses collaborative filtering methodologies, explainability and interpretability for recommender systems, prediction of consumer repurchasing cycles, reinforcement learning strategies, among other intriguing aspects.
eXplainable AI (XAI)
Explainable AI (XAI) bridges the gap between complex machine learning models and human interpretability, ensuring that AI-driven decisions are transparent and understandable. As AI systems become increasingly integrated into critical sectors like healthcare, finance, and law, it's essential for stakeholders to trust and validate the algorithms' outcomes. Moreover, XAI empowers users to identify potential biases, mistakes, or injustices, fostering more ethical, fair, and accountable AI applications.In DELTA Lab we develop novel XAI techniques with SOTA results.
Inflation is a key driver of economic policy decisions, and its accurate prediction can greatly influence fiscal stability and market confidence. Traditional economic models, while useful, often struggle to incorporate and process the sheer volume and complexity of data now available. Machine learning, with its ability to discern patterns and relationships in large datasets, holds great promise to enhance the accuracy and reliability of inflation forecasts. By developing new machine learning algorithms tailored for this task, we can potentially unlock deeper insights into economic trends, improve the precision of policy responses, and thereby contribute to sustainable economic growth and stability. This research is a collaboration with the Bank of Israel.
Machine Learning for Agricultural Improvements
We have a strong collaboration with the Volcani Institute Agricultural Research Organization. As part of this collaboration, DELTA Lab helps integrating advanced machine learning technologies for a wide area of agricultural applications ranging from post-harvest quality prediction, hyper-spectral seed scanning, and even maggot farming!
Machine Learning for Electronic Music Production
At DELTA, we are very interested in the problem of sound-matching for real-world synthesizers. Synthesizers are musical instruments that are widely used in electronic music production. Given an input sound, inferring the underlying synthesizer's parameters that will reproduce it is a difficult task known as sound-matching. Our research tackles the problem of automatic sound matching, which is otherwise performed manually by professional audio experts.
Dr. Noam Koenigstien
Head of DELTA Lab
Dr. Noam Koenigstein's academic journey began at the Technion – Israel Institute of Technology, where he graduated with honors in computer science for his B.Sc. degree. He furthered his education at Tel-Aviv University, earning a M.Sc. degree in Electrical Engineering, and subsequently, a Ph.D. from the same school focused on machine learning algorithms for recommender systems. In 2011, Dr. Koenigstein entered the industry as a member of Microsoft's Xbox Machine Learning research team, developing the algorithm for Xbox recommendations that reached millions of users globally. His trajectory within Microsoft culminated in him leading the recommendations research team for Microsoft’s Store. In 2017, he embarked on a new role as Senior VP Head of Data Science at Citi bank’s Israeli Innovation Lab, responsible for supervising all data science initiatives in the Israeli research center. In 2018, he transitioned back to academia, joining the Industrial Engineering department of Tel-Aviv University as a Senior Lecturer (Assistant Professor). Currently, he is the director of the DEep Learning Technologies and Applications (DELTA) Lab at Tel-Aviv University's Department for Industrial Engineering. Here, he mentors students in the application of machine learning algorithms to an array of real-world problems.
Dr. Liron Izhaki-Allerhand
Associate Researcher and Advisor
Dr. Liron Izhaki-Allerhand received his Ph.D. in the Department of Electrical Engineering in Tel Aviv University in 2014. He then assumed leadership roles in various renowned organizations, including a senior researcher at Microsoft. In recent years, he took a leadership role in Hour One, a company specializing in Generative AI, where he’s the Head of AI.
Dr. Veronika Bogina
Dr. Veronika Bogina completed her doctoral studies at Haifa University in 2022, specializing in the temporal aspects of user modeling within the Department of Information Systems. Before pursuing her Ph.D., she worked at IBM Haifa Labs in the Search Technologies group. Currently, she holds a postdoctoral research position at Tel Aviv University, where she is hosted by Noam Koenigstein.
Thesis: End-to-end Parameter Estimation for Sound Synthesis via Deep Networks
Thesis: Bayesian Model for CPI Estimation from Proxy Measurements
Thesis: Attention-Based Matrix Factorization Model For Explainable Recommendations
Thesis: Hierarchical Neural Models for Inflation Forecasting
Thesis: Machine Learning Models for Livestock Management
Thesis: Machine Learning Models for Fruit Preservation in Industrial Storage
Thesis: A Bayesian Model for CPI Estimation from Proxy Measurements
Thesis: Mind The Path: Path-Based Knowledge Graph with Neural Attention
Thesis: Neural Attentive Mixture Models
Thesis: Post-Harvest Quality Prediction in Oranges
Thesis: Attentive Item2Vec: Neural Attentive User Representations
Thesis: Automatic Synthesizer Sound Matching via Self-Supervised Synthesizer-Proxy Learning and Inference-Time Finetuning
Wolfson Building of Engineering Room 434
Department of Industrial Engineering
Tel Aviv University
P.O. Box 39040
Tel Aviv 6997801
We're here to listen and help, so feel free to contact us anytime.