[cs-talks] Upcoming CS Seminars: Data Seminar (Tues) + IVC (Tues) + BUSec Seminar (Wed) + Student Sem (Thurs)
fgreen1 at bu.edu
Mon Oct 26 10:59:02 EDT 2015
Word, Graph and Manifold Embedding from Markov Processes
Tatsu Hashimoto, MIT
Tuesday, October 27, 2015 at 11:30am in MCS 148
Abstract: Continuous vector representations of words and objects appear to carry surprisingly rich semantic content. In this paper, we advance both the conceptual and theoretical understanding of word embeddings in three ways. First, we ground embeddings in semantic spaces studied in cognitive-psychometric literature and introduce new evaluation tasks. Second, in contrast to prior work, we take metric recovery as the key object of study, unify existing algorithms as consistent metric recovery methods based on co-occurrence counts from simple Markov random walks, and propose a new recovery algorithm. Third, we generalize metric recovery to graphs and manifolds, relating co-occurrence counts on random walks in graphs and random processes on manifolds to the underlying metric to be recovered, thereby reconciling manifold estimation and embedding algorithms. We compare embedding algorithms across a range of tasks, from nonlinear dimensionality reduction to three semantic language tasks, including analogies, sequence completion, and classification.
A Deep Learning Approach
Kate Saenko, UMass Lowell
Tuesday, October 27, 2015 at 1pm in MCS 148
Abstract:I will describe several recent advances in automatic generation of natural language descriptions for video. Video description has important applications in human-robot interaction, video indexing, and describing movies for the blind. Real-world videos often have complex dynamics, but current methods are insensitive to temporal structure and do not allow both input (sequence of frames) and output (sequence of words) of variable length. I will describe a novel sequence-to-sequence neural network that learns to generate captions for brief videos. The model is trained on video-sentence pairs and is naturally able to learn the temporal structure of the sequence of frames as well as the sequence model of the generated sentences, i.e. a language model. To further handle the ambiguity over multiple objects and locations, the model incorporates convolutional networks with Multiple Instance Learning (MIL) to consider objects in different positions and at different scales simultaneously. The multi-scale multi-instance convolutional network is integrated with a sequence-to-sequence recurrent neural network to generate sentence descriptions based on the visual representation. This architecture is the first end-to-end trainable deep neural network that is capable of multi-scale region processing for video description. I will show results of captioning YouTube videos and Hollywood movies. This work is a joint project with Prof. Raymond Mooney's group at UT Austin and Prof. Trevor Darrell's group at UC Berkeley. Speaker
Bio: Kate Saenko is an Assistant Professor of Computer Science at the University of Massachusetts Lowell. She received her PhD from MIT, followed by postdoctoral work at UC Berkeley and Harvard. Her research spans the areas of computer vision, machine learning, speech recognition, and human-robot interfaces. Dr Saenko’s current research interests include domain adaptation for object recognition and joint modeling of language and vision.
Challenges in Blockchain Protocols
Ittay Eyal, Cornell
Wednesday, October 28, 2015 at 10am in MCS 180- Hariri Seminar Room
Abstract: Blockchain-based cryptocurrencies, based on Bitcoin, promise to become the infrastructure for pseudonymous online payments, cheap remittance, trustless digital asset exchange, and smart contracts. However, unique security aspects form challenges that must be overcome to realize this promise.
This talk starts with an overview of two results dealing with a key threat on blockchains, namely centralization, where a single principal comes to control the system. Then, it discusses the performance limits of blockchain protocols due to their security mechanism. Finally, it presents Bitcoin-NG, a novel Blockchain protocol that overcomes these limits.
Marcelo Coehlo, MIT (http://www.cmarcelo.com/)
Thursday, October 29, 2015 at 12pm in MCS 148
Description: Marcelo Coelho’s work is focused on programmable materials and digital fabrication technologies for products, installations, and crowd experiences. Spanning a wide range of media, processes, and scales, his work explores the boundaries between matter and information, fundamentally expanding and enhancing the ways in which we interact and communicate with one another.
Marcelo is currently the creative director at Marcelo Coelho Studio, where he develops innovative and experimental work. Prior to his studio practice, Marcelo Coelho received a Bachelor in Computation Arts with highest honors from Concordia University, and Doctorate in Media Arts and Sciences from the MIT Media Lab, where he continues to be a Research Affiliate. Marcelo’s creative work has been exhibited internationally, including places such as Ars Electronica, The Corcoran Gallery of Art, Tel Aviv Museum of Art, Design Miami/, The Creators Project, Riflemaker Gallery, Johnson Trading Gallery, MIT Museum, Waddesdon Manor, among others. His work can also be found in private collections including the Maxine and Stuart Frankel Foundation for Art, The Rothschild Collection and the Tech Museum of Innovation. His work has also won several awards, including two Prix Ars Electronica awards, VIDA 16.0 Award, the W Hotels Designer of the Future Award, and Honorable Mention from ID Magazine Design Review. His work has also been published and featured extensively in outlets such as the New York Times, Wired, Fast Company, Financial Times and the Colbert Report, in addition to books and academic peer-reviewed journals.
Marcelo Coelho currently spends his time between the USA and Brazil, developing new personal work, collaborations, and commissions.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the cs-talks