Videos

Intelligent Control of Crowdsourcing

Intelligent Control of Crowdsourcing

1 year ago
Crowd-sourcing labor markets (e.g., Amazon Mechanical Turk) are booming, because they enable rapid construction of complex workflows that seamlessly mix human computation with computer automation. Example applications range from photo tagging to audio-visual transcription and interlingual translation. Similarly, workflows on citizen science sites (e.g. GalaxyZoo) have allowed ordinary people to pool their effort and make […]
Improving Urban Public Education: Lessons from Charter Schools

Improving Urban Public Education: Lessons from Charter Schools

1 year ago
Charter schools represent one of the fastest growing, yet controversial innovations in education reform. In this talk, I will review several papers measuring urban school performance from a series of papers using data from Boston, New York City, Denver, and New Orleans from MIT’s School Effectiveness and Inequality Initiative. In addition to discussing the broader […]
An Algorithm for Precision Medicine

An Algorithm for Precision Medicine

1 year ago
President Obama recently launched the Precision Medicine Initiative, a confluence of efforts in data science, bioinformatics, systems biology and genomics. Precision medicine’s promise of “the right medicine to the right patient at the right time” is predicated on the assumption that a patient’s health data may be mapped directly to the “right medicine.” It is […]
VISC: Virtual Instruction Set Computing

VISC: Virtual Instruction Set Computing

1 year ago
Although many systems use virtual instruction sets to improve security, portability and productivity, two important classes of software — high performance applications and security-sensitive systems software — are almost universally compiled and shipped as native machine code. In the VISC project, we are exploring the security, reliability, and performance implications of shipping all such software […]
The Contextual Bandits Problem: A New, Fast, and Simple Algorithm

The Contextual Bandits Problem: A New, Fast, and Simple Algorithm

1 year ago
We study the general problem of how to learn through experience to make intelligent decisions. In this setting, called the contextual bandits problem, the learner must repeatedly decide which action to take in response to an observed context, and is then permitted to observe the received reward, but only for the chosen action. The goal […]
Deep consequences: Why syntax isn’t a thing; modeling language with neural nets

Deep consequences: Why syntax isn’t a thing; modeling language with neural nets

1 year ago
With the development of ‘deeper’ models of language processing, we can start to infer (in an more empirically sound way) the true principles, factors or structures that underline language. This is because, unlike many other approaches in NLP, deep language models (loosely) reflect the true situation in which humans learn language. Neural language models learn […]
DNN-Based Online Speech Enhancement Using Multitask Learning and Suppression Rule Estimation

DNN-Based Online Speech Enhancement Using Multitask Learning and Suppression Rule Estimation

1 year ago
Most of the currently available speech enhancement algorithms use a statistical signal processing approach to remove the noise component from observed signals. The performance of these algorithms is thus dependent on the statistical assumptions they make about speech and noise signals, which are often inaccurate. In this work, we consider machine learning as an alternative, […]
Research on Concert Hall Acoustics at Aalto University

Research on Concert Hall Acoustics at Aalto University

1 year ago
This talk will be about the research conducted in Virtual Acoustics team at Aalto University lead by professors Lauri Savioja and Tapio Lokki. It is related to the acoustic measurement and analysis of the concert halls, as well as their sensory evaluation. The ultimate goal is to understand how room acoustics modify the quality of […]
Program Synthesis from Refinement Types

Program Synthesis from Refinement Types

1 year ago
The key to scalable program synthesis is modular verification, which enables pruning inviable candidates for each component of the target program independently. In this talk I will present Synquid: a synthesizer that takes advantage of the modularity offered by refinement type checking to efficiently generate recursive functional programs that satisfy a given specification in the […]
Paradoxes of Openness and Distinction in the Sharing Economy

Paradoxes of Openness and Distinction in the Sharing Economy

1 year ago
Since the 1980s, Pierre Bourdieu’s influence in sociology has increased markedly, including on the study of consumption and economic life (Sallaz and Zavisca 2007). Bourdieu’s formulation of multiple types of capital (economic, cultural and social) and their role in producing and reproducing durable inequality has been highly productive in a variety of contexts. However, while […]
Change Dyslexia: Early Detection and Intervention at Large Scale

Change Dyslexia: Early Detection and Intervention at Large Scale

1 year ago
More than 10% of the population has dyslexia, and most are diagnosed only after they fail in school. My work is changing this through scalable early detection and tools that help people with dyslexia read and write better. To detect dyslexia, I am developing machine learning models that predict reading and writing difficulties by watching […]
Big Data and Bayesian Nonparametrics

Big Data and Bayesian Nonparametrics

1 year ago
Big Data is often characterized by large sample sizes, high dimensions, and strange variable distributions. For example, an e-commerce website has 10-100s million observations weekly on a huge number of variables with density spikes at zero and elsewhere and very fat tails. These properties — big and strange — beg for nonparametric analysis. We revisit […]