Automatic speech recognition (ASR) has been a grand challenge machine learning problem for decades. Our ongoing research in this area examines the use of deep learning models for distant and noisy recording conditions, multilingual, and low-resource scenarios.
We study the fundamentals of Bayesian optimization and develop efficient Bayesian optimization methods for global optimization of expensive black-box functions originated from a range of different applications.
BlueDBM is an architecture of computer clusters consisting of fast distributed flash storage and in-storage accelerators, which often outperforms larger and more expensive clusters in applications such as graph analytics.
Our main goal is to automatically search for relevant answers among many responses provided for a given question (Answer Selection), and search for relevant questions to reuse their existing answers (Question Retrieval).
We aim to understand theory and applications of diversity-inducing probabilities (and, more generally, "negative dependence") in machine learning, and develop fast algorithms based on their mathematical properties.
Our main goal is to develop fact checking algorithms that can assess the credibility of claims mentioned in the textual statements and provide interpretable valid evidence that explains why a certain claim is considered as factually true or fake.
Data often has geometric structure which can enable better inference; this project aims to scale up geometry-aware techniques for use in machine learning settings with lots of data, so that this structure may be utilized in practice.
The MOOC Learner Project provides learning scientists, instructional designers and online education specialists with open source software that enables them to efficiently extract teaching and learning insights from the data collected when students learn on the edX or open edX platform.
Generation of sequential data involves multiple factors operating at different temporal scales. Take natural speech for example, the speaker identity tends to be consistent within an utterance, while the phonetic content changes from frame to frame. By explicitly modeling such hierarchical generative process under a probabilistic framework, we proposed a model that learns to factorizes sequence-level factors and sub-sequence-level factors into different sets of representations without any supervision.
All humans process vast quantities of unannotated speech and manage to learn phonetic inventories, word boundaries, etc., and can use these abilities to acquire new word. Why can't ASR technology have similar capabilities? Our goal in this research project is to build speech technology using unannotated speech corpora.
Last week MIT’s Institute for Foundations of Data Science (MIFODS) held an interdisciplinary workshop aimed at tackling the underlying theory behind deep learning. Led by MIT professor Aleksander Madry, the event focused on a number of research discussions at the intersection of math, statistics, and theoretical computer science.
Last month, three MIT materials scientists and their colleagues published a paper describing a new artificial-intelligence system that can pore through scientific papers and extract “recipes” for producing particular types of materials.
Neural networks, which learn to perform computational tasks by analyzing huge sets of training data, have been responsible for the most impressive recent advances in artificial intelligence, including speech-recognition and automatic-translation systems.
Artificial intelligence (AI) in the form of “neural networks” are increasingly used in technologies like self-driving cars to be able to see and recognize objects. Such systems could even help with tasks like identifying explosives in airport security lines.
We live in the age of big data, but most of that data is “sparse.” Imagine, for instance, a massive table that mapped all of Amazon’s customers against all of its products, with a “1” for each product a given customer bought and a “0” otherwise. The table would be mostly zeroes.
Every year 40,000 women die from breast cancer in the U.S. alone. When cancers are found early, they can often be cured. Mammograms are the best test available, but they’re still imperfect and often result in false positive results that can lead to unnecessary biopsies and surgeries.
Most modern websites store data in databases, and since database queries are relatively slow, most sites also maintain so-called cache servers, which list the results of common queries for faster access. A data center for a major web service such as Google or Facebook might have as many as 1,000 servers dedicated just to caching.
Laparoscopy is a surgical technique in which a fiber-optic camera is inserted into a patient’s abdominal cavity to provide a video feed that guides the surgeon through a minimally invasive procedure. Laparoscopic surgeries can take hours, and the video generated by the camera — the laparoscope — is often recorded. Those recordings contain a wealth of information that could be useful for training both medical providers and computer systems that would aid with surgery, but because reviewing them is so time consuming, they mostly sit idle.
This week the Association for Computer Machinery presented CSAIL principal investigator Daniel Jackson with the 2017 ACM SIGSOFT Outstanding Research Award for his pioneering work in software engineering. (This fall he also received the ACM SIGSOFT Impact Paper Award for his research method for finding bugs in code.)An EECS professor and associate director of CSAIL, Jackson was given the Outstanding Research Award for his “foundational contributions to software modeling, the creation of the modeling language Alloy, and the development of a widely used tool supporting model verification.”
Speech recognition systems, such as those that convert speech to text on cellphones, are generally the result of machine learning. A computer pores through thousands or even millions of audio files and their transcriptions, and learns which acoustic features correspond to which typed words.But transcribing recordings is costly, time-consuming work, which has limited speech recognition to a small subset of languages spoken in wealthy nations.
Every language has its own collection of phonemes, or the basic phonetic units from which spoken words are composed. Depending on how you count, English has somewhere between 35 and 45. Knowing a language’s phonemes can make it much easier for automated systems to learn to interpret speech.In the 2015 volume of Transactions of the Association for Computational Linguistics, CSAIL researchers describe a new machine-learning system that, like several systems before it, can learn to distinguish spoken words. But unlike its predecessors, it can also learn to distinguish lower-level phonetic units, such as syllables and phonemes.