Monthly Archives: October 2016

Xue in MLFL Thursday Nov. 3 at noon

Tianfan Xue will present “Visual Dynamics Probabilistic Future Frame Synthesis Via Cross Convolutional Networks” in the Machine Learning and Friends Lunch this Thursday at noon. Details, abstract and short bio are below.

What is it? A gathering of students/faculty/staff with broad interest in the methods and applications of machine learning.
When is it? Thursdays 12:00pm to 1:00pm, unless otherwise noted. Arrive at 11:45 to get pizza.
Where is it? CS150
Who is invited? Everyone is welcome.
Is there food? Yes! Pizza is provided.

Visual Dynamics Probabilistic Future Frame Synthesis Via Cross Convolutional Networks

Abstract:
We study the problem of synthesizing a number of likely future frames from a single input image. In contrast to traditional methods, which have tackled this problem in a deterministic or non-parametric way, we propose to model future frames in a probabilistic manner. Our probabilistic model makes it possible for us to sample and synthesize many possible future frames from a single input image. To synthesize realistic movement of objects, we propose a novel network structure, namely a Cross Convolutional Network; this network encodes image and motion information as feature maps and convolutional kernels, respectively. In experiments, our model performs well on synthetic data, such as 2D shapes and animated game sprites, as well as on real-world video frames. Please refer to our website for more details: http://visualdynamics.csail.mit.edu/.

Short bio:
Tianfan Xue is currently a fifth-year Ph.D. student in MIT CSAIL, working with William T. Freeman. Before that, he received his B.E. degree from Tsinghua Universtiy, and M.Phil. degree from The Chinese University of Hong Kong. His research interests include computer vision, image processing, and machine learning. Specifically, he is interested in motion estimation and image and video processing based on the motion information.

Gow in Linguistics 3:30 Fri. Nov. 4

David Gow (Cognitive/Behavioral Neurology Group, Massachusetts General
Hospital), will present “Inference, phonology and the brain: What Granger analysis can tell us about the sources of phonological structure” Friday November 4, 2106, at 3:30 PM, in ILC N400. An abstract follows.

Abstract: Speech perception reflects the lawful phonological patterning
of language. This has been explained reasonably well through vastly
different approaches involving generative phonological rules and
constraints, statistical inference, and interactive associative mapping
processes. This three-way distinction can be distilled to different
accounts of the functional architecture of language processing.
Unfortunately, claims about functional architecture (e.g. modularity
versus interactivity) have proven notoriously hard to falsify using
traditional behavioral and BOLD imaging techniques. In this talk I will
introduce the use of Kalman-filter enabled Granger causation analysis of
MR-constrained MEG/EEG data as a powerful new tool to discover
functional architecture. By identifying the pattern of directed
influences between functionally interpretable brain regions during task
performance, this technique provides an entirely data-driven approach
for discovering functional architecture. Using this approach, I will
present data that challenges both statistical and rule/constraint
accounts of phonotactic influences on perception, and suggests that
phonotactic phenomena are the result of top-down lexical influences on
speech perception.

Potter in Cognitive Brown Bag Weds. Nov. 2 at noon

Time: 12:00pm to 1:15pm Wednesday Nov. 2. Location:  Tobin 521B. All are welcome!

Kevin Potter (University of Massachusetts)

Title:
Testing a perceptual fluency/disfluency model of priming with a model of response time and choice

Abstract:
With immediate repetition priming of forced choice perceptual identification, short prime durations produce positive priming (i.e., higher accuracy when the target is primed, but lower accuracy when the foil is primed). In contrast, long prime durations reverse this pattern. The dynamic time course of this transition from positive to negative priming is well explained by the nROUSE model of Huber and O’Reilly (2003), which includes neural habituation. This model assumes that the speed of perceptual identification is used to decide which choice word was seen most recently as the briefly flashed target. Thus, short duration primes induce faster identification (perceptual fluency) for the primed choice and a bias for the primed alternative whereas long duration primes induce slower identification (perceptual disfluency) for the primed choice and a bias against the primed alternative. This account makes specific predictions regarding perceptual identification latencies, and yet a test of these predictions is difficult with forced choice testing, which reflects a comparison decision process. To address this limitation, we collected forced-choice and single-item same-different responses in the same priming paradigm. We then applied a diffusion-race model to the data, transforming the response time and choice data into ‘observed’ drift rate parameters (i.e., the rate of evidence accumulation). Remarkably, the drift rates were inversely proportional to the identification latencies of the nROUSE model even though each model was applied independently to the data and even though the nROUSE model was only applied to the accuracy data. This convergence of the models confirms key predictions of the nROUSE model regarding perceptual fluency and disfluency.

Jackendoff in Linguistics 3:30 Fri. Oct. 28

Ray Jackendoff will present a colloquium in the Department of Linguistics this Friday (October 28). Place: N400. Time: 3:30 PM.

Ray Jackendoff (Tufts University) and Jenny Audring (University of Leiden): Morphology in the Mental Lexicon.

We explore a theory of morphology grounded in the outlook of the Parallel Architecture (PA, Jackendoff 2002), drawing in large part on Construction Morphology (Booij 2010). The fundamental goal is to describe what a speaker stores and in what form, and to describe how this knowledge is put to use in constructing novel utterances. A basic tenet of PA is that linguistic structure is built out of independent phonological, syntactic, and semantic/conceptual structures, plus explicit interfaces that relate the three structures, often in many-to-many fashion.

Within this outlook, morphology emerges as the grammar of word-sized pieces of structure and their constituents, comprising morphosyntax and its interfaces to word phonology, lexical semantics, and phrasal syntax. Canonical morphology features a straightforward mapping among these components; irregular morphology is predominantly a matter of noncanonical mapping between constituents of morphosyntax and phonology.

As in Construction Grammar and Construction Morphology, PA encodes rules of grammar as schemas: pieces of linguistic structure that contain variables, but which are otherwise in the same format as words – in other words, the grammar is part of the lexicon. Novel utterances are constructed by instantiating variables in schemas through Unification. A compatible morphological theory must likewise state morphological patterns in terms of declarative schemas rather than procedural or realizational rules.

Non-productive morphological patterns can be described in terms of schemas that are formally parallel to those for productive patterns. However, they do not encode affordances for building new structures online; rather, they motivate relations among items stored in the lexicon. Productive schemas can be used in this way as well, in addition to their standard use in building novel structures; hence they can be thought of as schemas that have “gone viral.” Interestingly, this classification proves useful also for extending syntactic schemas to idioms and other fixed expressions.

This raises the question of how lexical relations are to be expressed. Beginning with the well-known mechanism of inheritance, we show that inheritance should be cashed out, not in terms of minimizing the number of symbols in the lexicon, but in terms of increased redundancy (or lower entropy). We propose a generalization of inheritance to include lexical relations that are nondirectional and symmetrical, and we develop a notation that pinpoints the regions of commonality between pairs of words, between words and schemas, and between pairs of schemas.

We conclude that linguistic theory should be concerned with relations among lexical items, from productive to marginal, at least as much as with the online construction of novel forms. We further conclude that the lexicon is richly textured, in a fashion that invites comparison with other domains of human knowledge.

Boroditsky at Smith College Thurs. Oct. 27 at 4:30

Lera Boroditsky will present “How the languages we speak shape the ways we think” on Thursday, October 27, 4:30 PM, McConnell Foyer 103, Smith College.

Lera Boroditsky is an Associate Professor of Cognitive Science at UCSD and Editor in Chief of Frontiers in Cultural Psychology. She previously served on the faculty at MIT and at Stanford. Her research is on the relationships between mind, world, and language (or how humans get so smart). She has been named one of 25 Visionaries changing the world by the Utne Reader, and is also a Searle Scholar, a McDonnell scholar, recipient of an NSF Career award, and an APA Distinguished Scientist lecturer.

Criss in Cognitive Brown Bag Weds. Oct. 26th at noon

Time: 12:00pm to 1:15pm Wednesday Oct. 26. Location:  Tobin 521B.

Amy Criss (Syracuse University)

http://memolab.syr.edu/

Title: Memory across Tasks, Items, and Individuals

Memory researchers have spent the past several decades drilling down into memory. For example, many empirical investigations and theoretical developments focus on a single task and/or a single effect. We propose that the field is well-positioned to benefit from zooming back out. Here we present one attempt to do so. 462 participants completed 5 memory tasks with a fixed set of words that varied on many dimensions. We extract factors that are shared across memory tasks and we identify properties of items that support memory.  Finally, we sketch the beginnings of a conceptual model based on these findings.

Ng in CSSI, Weds. Oct. 19 at 11 am

In lieu of the Friday Computational Social Science Institute seminar this week there is a collaborative talk with the UMass Information Technology Policy seminar series. The talk will take place in CS140, Wednesday October 19  11AM.

Jason Q. Ng, Data Analyst Tumblr; Research Fellow, Citizen Lab
A data-driven approach to researching censorship and sensitive conversations on social media 

Like all nations, China has been profoundly affected by the emergence of the Internet, particularly new forms of social media which allow individuals themselves to be independent broadcasters of news. However the rise of “We Media” has also led to a corresponding rise in the filtering and blocking of online content in China. Identifying and explaining these disruptions comes with a host of challenges for researchers–ranging from technical ones like developing methodologies for tracking online censorship to non-technical ones like even defining what online censorship is. 

In this talk, we’ll look at a number of different ways online censorship can be defined as well as various data-driven techniques for revealing its occurrence in social media–as well as the various ways social media companies attempt to mask or justify it. However, just as important as identifying the mechanisms for how censorship is implemented is trying to understand the motivations for such behavior. Knowing both how and why online censorship occurs is key for not only academic researchers who hope to better understand content moderation and filtering practices, but it is essential information for the activists, journalists, and advocates who utilize such findings in their work.

About the speaker: Jason Q. Ng is currently a Research Fellow at the University of Toronto’s Citizen Lab, Data Analyst at Tumblr, and author of Blocked on Weibo, a book on Chinese social media. He is also a research consultant at China Digital Times where he develops censorship monitoring tools and teaches a digital activism course at Columbia SIPA. His writing and research projects can be found at www.jasonqng.com.

Weidemann in Cognitive Bag Lunch Weds. Oct. 19

Time: 12:00pm to 1:15pm Wednesday Oct. 19. Location:  Tobin 521B. Christoph Weidemann (Swansea University) http://cogsci.info/ “The dynamics of recognition memory: Insights from response times and brain activity”

Abstract: Decades of research on recognition memory has given us great insights into the underlying processes and features of stimuli and tasks that improve or impair performance. However relatively little is known about how the signal on which memory decisions are based evolves across a recognition memory trial. Despite well-known issues with introspective judgments, most studies attempting to measure “memory strength” rely on confidence ratings at or after the classification of an item as “old” or “new”. I will show that much of information in confidence ratings can be obtained from response times. Furthermore, I will show how evidence distinguishing targets from lures steadily accumulates during the recognition test by applying machine learning techniques to EEG data. Implications of this approach for the study of recognition memory and other classification tasks are discussed.

Utts at Smith College Tues. Oct. 18th at 4 pm

Understanding p-values and the Controversy Surrounding Them

SPEAKER:               Jessica Utts

President of the American Statistical Association

Professor of Statistics, University of California at Irvine

DATE:                       Tuesday, October 18, 2016

TIME:                       Tea at 4 o’clock

Talk at 4:30 pm

PLACE:                    Tea in McConnell Foyer (Smith College)

Talk in McConnell 103

ABSTRACT:

Most researchers and journals rely heavily on p-values for determining whether the results of a study are worthy of publication. But recently p-values have come under attack, and one social science journal has gone as far as banning their use for papers submitted to the journal. These developments led the American Statistical Association (ASA) to release a statement titled “Statement on Statistical Significance and P-values” with six principles underlying the proper use and interpretation of p-values and statistical significance. In this talk I will present the ASA’s six principles and discuss what p-values really measure, some pitfalls related to their use, and what steps you can take to make sure your use of them is appropriate.

SPONSORED BY:                   The Boston Chapter of the American Statistical Association

Smith College

The Five-Colleges Statistics Program

Fisher in Distinguished Faculty Lecture Tues. Oct. 18th at 4 p.m.

The campus community is invited to attend the first 2016-2017 Distinguished Faculty Lecture on Tuesday, October 18, at 4 p.m. in the Bernie Dallas Room, Goodell Building.

Professor Donald Fisher, Department of Mechanical and Industrial Engineering, will present “The Eyes Have It: A Window into the Mind.”

A common thread links long-standing questions in cognitive science, gerontology, transportation, and health care. Cognitive scientists want to know whether we really need to put aside our cell phones while driving. Gerontologists want to know whether some of our cognitive processes can be spared as we age. Transportation specialists want to know why younger and older drivers crash so often. Health care officials want to know why some 100,000 deaths each year are attributed to medical errors. In seeking answers to these and other questions, we need to understand how to use information that arrives at the eyes in order to shed light on the latent cognitive processes that govern performance. This lecture will show how such an understanding can help answer these questions and why such an understanding is essential in a time when machines (e.g., autonomous vehicles) are seemingly making such understanding ever less necessary.

At the conclusion of the lecture, Professor Fisher will be presented with the Chancellor’s Medal, the highest honor bestowed to faculty by the campus.

This lecture is free and open to the public, with a reception immediately following the lecture.

Additional information about the 2016-2017 Distinguished Faculty Lecture Series is available at http://www.umass.edu/faculty-lecture.