Bloomberg Data Science Research Grant Program

Bloomberg invites academic researchers worldwide to apply for unrestricted gifts that support research in data science, typically focusing on natural language processing, information retrieval, machine learning, and data mining. We also invite proposals for creation of, or contributions to, open source software used for data science.

Call for Applications

Faculty members, research scientists, and post-doctoral fellows at universities worldwide are eligible to serve as Principal Investigators (PIs). A PI can lead one proposal per cycle, but can serve as co-PI on other proposals.

Applications should be submitted here in PDF format. We prefer a single file containing all materials, named after the first PI’s last name and first name. Please note that Bloomberg cannot accept any proposal containing confidential or proprietary information.

Applications must be submitted by March 15, 2018. Grant decisions will be communicated by May 15, 2018.

Criteria

We look for technical merit, novelty, and potential for impact. Good proposals present a data science problem, a specific idea, and argue that this idea has the potential to succeed. Proposals to investigate an area without a plan of attack are less likely to be funded. We also prefer to fund research that is not easily funded otherwise.

If the proposal depends on other funding or non-public data, the proposal should state whether it has been secured.

Proposals are reviewed by a committee of Bloomberg employees from various areas of data science. The bulk of the proposal should be directed at a technical data science reviewer, but we also ask that it include one to two paragraphs making the case to a technical non-expert.

Bloomberg employees in your research community can provide feedback before you submit a proposal. Contacting employees is not required for funding, but it can help. All funded projects will be assigned a contact at Bloomberg.

Open Research

Participants are strongly encouraged to openly release papers, data, and programs that result from their research. Proposals should state the PI’s intentions, which will be considered during the review. However, because this program provides unrestricted gifts, publication of research results is not a contractual requirement. PIs previously funded by Bloomberg should indicate whether their results are open.

Structure

The proposal should contain four sections: title, technical details, budget, and data policy. The maximum length of a proposal is three pages including references. Each PI should submit his/her CV along with the proposal. If any of the PIs have received funding from Bloomberg in the past, they should include an additional half page summary of the results (total three and a half pages).

Application Details

Applicants must submit only one proposal for the 2018-2019 academic year, and this proposal must be submitted to only one grant program; cross-submission to the Bloomberg Ph.D. Fellowship and the Research Grant Program will result in disqualification.

Title

Please provide names, contacts, and affiliations of the PIs along with the proposal title. If any Bloomberg employees were consulted in preparing the proposal, please provide their names as well.

Technical Details

The bulk of the proposal should be directed at a technical data science audience, but we also ask that it include one or two paragraphs making the case to a non-expert. A typical proposal will contain:

Abstract, problem statement, and the PI’s ideas
Description of the proposed work and expected results
Relevant references

Budget

Along with the total amount (in USD) requested in the proposal, it should also provide a brief breakdown of the direct costs, including: student/post-doc stipend, tuition fees, conference travel, and hardware. Typically, we fund one student for one year, which is often 40,000 to 70,000 USD.

Indirect costs such as administrative overhead will not be funded.

Data Policy

Please briefly state the means by which the results produced by this funding will be disseminated (e.g. open source software, open access journals, conferences, and/or presentations).

Contact Information

If you have questions about the program, or how to construct your proposal, you can contact the grant program review team via email at rdml@bloomberg.net.

Additional Information

Grant awardees will be given access to a Bloomberg Professional Service (Terminal) account upon request. At least one PI from each successful proposal will be expected to present their (possibly partial) results and meet our data science researchers at Bloomberg’s New York or London offices. Costs for this travel will be borne by Bloomberg and are separate from the grant award. Grant awardees may separately apply to spend their sabbatical year or summers at Bloomberg. We also encourage students working under the grant awardees to apply for summer internships.

Previous Grant Recipients

Round 5 – July 2018

Neural Information Retrieval with Limited Data, Bruce Croft (University of Massachusetts Amherst)

Interactive Explorative Summarization: Closing the Summarization Gap, Ido Dagan (Bar Ilan University)

Representing Knowledge by Learning to Link, Kevin Gimpel and Karl Stratos (Toyota Technological Institute [TTI] at Chicago)

Cost-effective Learning for Complex Crowdsourcing Tasks, Xi Chen (NYU Stern School of Business)

Task Oriented Information Interaction Systems for Proactive Task Assistance Support, Emine Yilmaz (University College London)

Round 4 – April 2017

Combining structured knowledge and big data for coreference resolution, Greg Durrett (University of Texas – Austin)

Question answering and reasoning in multimodal data, Hannaneh Hajishirzi (University of Washington)

Entity salience via sophisticated syntactic and semantic features, Paolo Ferragina (Universita di Pisa, Italy)

Counterfactual learning with log data, Thorsten Joachims (Cornell University)

Learning hidden semantics by machine reading using entailment graphs, Mark Steedman (University of Edinburgh)

Deep explanation learning for knowledge graph relations, Maarten de Rijke (University of Amsterdam)

Dynamic word embeddings and applications in analysis of real-world discourses, Simon Preston, Karthik Bharath, Yves van Gennip (University of Nottingham) and Michaela Mahlberg (University of Birmingham)

Coarse-to-fine neural attention and generation with applications to document analysis, Alexander Rush (Harvard University)

Round 3 – April 2016

Spectral Learning with Prior Information with Applications to Topic Models, Daniel Hsu (Columbia University) and Kamalika Chaudhry (University of California, San Diego)

Dynamic Interpretability in Machine Learning, Yisong Yue (California Institute of Technology)

Latent-Variable Spectral Learning Kernelization for NLP, Shay Cohen (University of Edinburgh)

Online clustering of time-sensitive data, Stephen Becker (University of Colorado at Boulder)

Character-level neural network sequence models for varied text named entity recognition, Christopher Manning (Stanford University)

What’s The Angle? Disentangling Perspectives from Content in the News, Noah Smith (University of Washington), Amber Boydstun (University of California, Davis), Philip Resnik (University of Maryland), Justin Gross (University of Massachusetts, Amherst)

Multimodal Event Summarization, Mohit Bansal (TTI-Chicago, UNC-Chapel Hill)

Contextual Entity Recommendation, Maarten de Rijke (University of Amsterdam)

Round 2 – October 2015

Deep Topic Models, Professor Alexander Smola and Professor Chris Dyer, Carnegie Mellon University

Distributed Local Learning via Random Forests, Professor Ameet Talwalkar, University of California at Los Angeles

Establishing Trust in Tweets, Professor Mark Dredze, Johns Hopkins University

Report Linking, Professor Benjamin Van Durme, Johns Hopkins University

Coherent Multi-Document Summarization, Professor Mausam, Indian Institute of Technology (IIT), Delhi

Round 1 – April 2015

Scalable Probabilistic Deep Learning, Professor Jinwoo Shin, KAIST, South Korea

Algorithms for Offline, Online and Stochastic Clustering, Professor Viswanath Nagarajan, University of Michigan

Latent-Variable Learning for Transition-Based Parsing, Professor Shay Cohen, University of Edinburgh and Giorgio Satta, University of Padua