News and Events

AI EVENTS

AI Horizons Seminar Series
Tuesday, 12 November 2019 1 PM EST / 1800 GMT https://ibm.webex.com/join/aihn

Open Source Project - https://github.com/ibm/lale

Abstract:
Machine-learning automation tools, ranging from humble grid-search to hyperopt, auto-sklearn, and TPOT, help explore large search spaces of possible pipelines. Unfortunately, each of these tools has a different syntax for specifying its search space, and the search spaces are not necessarily consistent with error checks and documentation of the searchable base components. Lale is an open source Python library that proposes using types (such as enum, float, or dictionary) both for checking the correctness of, and for automatically searching over, hyperparameters and pipeline configurations. Using types for both of these purposes guarantees consistency. Lale resembles scikit-learn but provides better automation and correctness checks. It helps users leverage automation with minimal code while remaining in control of their work.

Bio:
Kiran Kate is a senior software engineer at IBM Research. Kiran has worked on machine learning for 10+ years and has experience in building ML applications such as statistical machine translation, recommendation systems, and ML for manufacturing industry. Kiran did her Master’s in Computer Science from Indian Institute of Technology Madras, India with a gold medal for highest CGPA.

past seminars: https://ibm.biz/aihn-recordings

AI Horizons Seminar Series
Monday, 21 October 2019 11 AM EDT / 1500 GMT https://ibm.webex.com/join/aihn

ACL 2019 - https://www.aclweb.org/anthology/P19-1473

Abstract:
Semantic parsing over multiple knowledge bases enables a parser to exploit structural similarities of programs across the multiple domains. However, the fundamental challenge lies in obtaining high-quality annotations of (utterance, program) pairs across various domains needed for training such models. To overcome this, we propose a novel framework to build a unified multi-domain enabled semantic parser trained only with weak supervision (denotations). Weakly supervised training is particularly arduous as the program search space grows exponentially in a multi-domain setting. To solve this, we incorporate a multipolicy distillation mechanism in which we first train domain-specific semantic parsers (teachers) using weak supervision in the absence of the ground truth programs, followed by training a single unified parser (student) from the domain specific policies obtained from these teachers. The resultant semantic parser is not only compact but also generalizes better, and generates more accurate programs. It further does not require the user to provide a domain label while querying. On the standard OVERNIGHT dataset (containing multiple domains), we demonstrate that the proposed model improves performance by 20% in terms of denotation accuracy in comparison to baseline techniques.

Bio:
Ayushi Dalmia is working as a research engineer in the AI for Interaction Department at IBM Research. Currently, she is contributing to the NLQ Project. Prior to this she worked on AI for Fashion. Before joining IBM Research, Ayushi finished her master's from IIIT Hyderabad where her thesis was on Outlier Detection in Graphs.  Her research interests include natural language processing, machine learning, deep learning and network analysis. Her work has appeared at venues such as ACL, KDD, CSUR, ICDM. 

past seminars: https://ibm.biz/aihn-recordings
 

AI Horizons Seminar Series
Thursday, 10 October 2019 3pm EDT / 1900 GMT https://ibm.webex.com/join/aihn

ICCV 2019 - https://arxiv.org/abs/1811.08815

Abstract:
Fine-grained action detection is an important task with numerous applications in robotics and human-computer interaction. Existing methods typically utilize a two-stage approach including extraction of local spatio-temporal features followed by temporal modeling to capture long-term dependencies. While most recent papers have focused on the latter (long-temporal modeling), here, we focus on producing features capable of modeling fine-grained motion more efficiently. We propose a novel locally-consistent deformable convolution, which utilizes the change in receptive fields and enforces a local coherency constraint to capture motion information effectively. Our model jointly learns spatio-temporal features (instead of using independent spatial and temporal streams). The temporal component is learned from the feature space instead of pixel space, e.g. optical flow. The produced features can be flexibly used in conjunction with other long-temporal modeling networks, e.g. ST-CNN, DilatedTCN, and ED-TCN. Overall, our proposed approach robustly outperforms the original long-temporal models on two fine-grained action datasets: 50 Salads and GTEA, achieving F1 scores of 80.22% and 75.39% respectively.

Bio:
Khoi-Nguyen is a Ph.D. Candidate at the University of Illinois at Urbana-Champaign, Department of Electrical and Computer Engineering. He is working with Prof. Minh N. Do, in Coordinated Science Laboratory's Computational Imaging Group and IBM's Center for Cognitive Computing Systems Research (C3SR). His research interests include Action Detection and Recognition, Machine Learning, Computer Vision, and Signal Processing.

past seminars: https://ibm.biz/aihn-recordings
 

AI Horizons Seminar Series
Monday, 9 September 2019 3pm EDT / 1900 GMT https://ibm.webex.com/join/aihn

ACL 2019 - https://arxiv.org/abs/1906.08042

Abstract:
Entity resolution (ER) is the task of identifying different representations of the same real-world entities across databases. It is a key step for knowledge base creation and text mining. Recent adaptation of deep learning methods for ER mitigates the need for dataset-specific feature engineering by constructing distributed representations of entity records. While these methods achieve state-of-the-art performance over benchmark data, they require large amounts of labeled data, which are typically unavailable in realistic ER applications. In this paper, we develop a deep learning-based method that targets low-resource settings for ER through a novel combination of transfer learning and active learning. We design an architecture that allows us to learn a transferable model from a high-resource setting to a low-resource one. To further adapt to the target dataset, we incorporate active learning that carefully selects a few informative examples to fine-tune the transferred model. Empirical evaluation demonstrates that our method achieves comparable, if not better, performance compared to state-of-the-art learning-based methods while using an order of magnitude fewer labels.

Bio: 
Jungo Kasai is a second-year PhD student at the Paul G. Allen School of Computer Science & Engineering of the University of Washington, Seattle, advised by Noah A. Smith. He works on natural language processing and machine learning. His papers have been accepted to conferences, such as ACL, NAACL, and EMNLP. His research interests include representation learning for multilingual natural language processing, structured predictions, and syntactic parsing.

past seminars: https://ibm.biz/aihn-recordings

AI Research Week
September 16 – 20, 2019

AI Research Week is five days of innovation, inspiration, and insights featuring notable speakers, panels, workshops, networking, and mentorship from leading figures in AI research.
View the full schedule

News from Rensselaer

TROY, N.Y. —Machine learning has the potential to vastly advance medical imaging, particularly computerized tomography (CT) scanning, by reducing radiation exposure and improving image quality.

Those new research findings were just published in Nature Machine Intelligence by engineers at Rensselaer Polytechnic Institute and radiologists at Massachusetts General Hospital and Harvard Medical School.

TROY, N.Y. — A wide-eyed, soft-spoken robot named Pepper motors around the Intelligent Systems Lab at Rensselaer Polytechnic Institute. One of the researchers tests Pepper, making various gestures as the robot accurately describes what he’s doing. When he crosses his arms, the robot identifies from his body language that something is off. 

“Hey, be friendly to me,” Pepper says.

TROY, N.Y. — Generating comprehensive molecular images of organs and tumors in living organisms can be performed at ultra-fast speed using a new deep learning approach to image reconstruction developed by researchers at Rensselaer Polytechnic Institute.

The research team’s new technique has the potential to vastly improve the quality and speed of imaging in live subjects and was the focus of an article recently published in Light: Science and Applications, a Nature journal.

TROY, N.Y. — Researchers at Rensselaer Polytechnic Institute who developed a blood test to help diagnose autism spectrum disorder have now successfully applied their distinctive big data-based approach to evaluating possible treatments.

Past AIHN Seminars Playlist