At Facebook, we apply statistics to generate insights and to improve decision making for a business that touches the lives of billions of people across the globe. Making advances in statistical methodologies helps us make the best decisions for our community, products, and infrastructure. For example, we develop prevalence algorithms that can be used for rare events and bias correction, as well as accounting for the noise of labeled data (metrics included in the Facebook Transparency Report). We also design novel inference algorithms that leverage the complex social processes to detect violating behaviors online and protect the community. This includes the Deep Entity Classifier (DEC) or the Temporal Interaction EmbeddingS (TIES) models.
We are following up on the 2019 Statistics for Improving Insights and Decisions research awards to foster further innovation in this area and to deepen our collaborations with academia. External researchers can submit proposals to address challenges in applied statistics that have direct applications for producing more effective insights and decisions for data scientists and researchers. Facebook has a large, active and diverse community of practitioners, so we are interested in a varied set of statistical topics, including but not limited to experimentation, forecasting, predictive modeling, survey and ground truth modeling and sampling.
The following are some specific examples of topics that are important to Facebook, but we will gladly review proposals for areas that are not listed below.
- Learning and evaluation under uncertainty. To protect people online and provide them with a meaningful experience, Facebook develops predictive models, which typically require training and evaluation data. Obtaining this data is oftentimes resource intensive, e.g., manual labeling, surveys and product interactions. It is therefore subject to noise and biases. We are interested in practical methodologies that address these challenges, estimate various biases, reduce the resources used for obtaining labels and produce calibrated estimations and predictions across cohorts of varying sizes.
- Statistical models of complex social processes. Facebook’s products help connect billions of people, and we often think of our products and systems as time-varying networks at scale. As these networks of connections evolve, various social processes also unfold on top of them: content spreads, social groups form and dissolve, people leverage their networks to organize events and support charitable causes, etc. Statistical models, both of the evolution of connection networks and of social processes on top of those networks, provide important input into Facebook’s efforts to design better products and to build safer and more meaningful communities. We invite proposals around the development, inference, and validation of such statistical models.
- Causal inference with observational data. Often researchers at Facebook would like to answer causal questions even when it is not possible to conduct product tests. For instance, we may want to measure the effects of external events or understand the potential causes for anomalies that we observe in our data. Proposals in this area should improve our ability to suggest potential hypotheses for interesting phenomena or to credibly estimate the effects of known causes. Another application of interest is enabling Facebook to predict how key app performance and reliability metrics will change when upgrades are rolled out to the entire user base based on the treatment effect observed on a selected population during the test phase of the app. We are also interested in the related field of survey methodology – dealing with non response or missing data.
- Efficient sampling and prevalence measurement. An important problem for Facebook is monitoring and measuring how well we do at removing content that violates our Community Standards from the platform, such as hate speech, spam and nudity. We are interested in methods that can reduce the latency, computational cost, bias and variance with which we measure the prevalence or reach of entities (posts, groups, users, …), in particular when this prevalence is low.
- Design and analysis of experiments. Facebook uses randomized experiments to measure the benefits of the improvements we make to our products. We seek to maximize what we learn from these experiments by improving how they are designed and analyzed. We are interested in methodologies which allow us to extend or enhance the standard experimentation framework: variance reduction; measuring heterogenous or time-varying effects; estimating effects of many-valued or continuous-valued treatments; aggregating information across multiple related experiments; and correcting for selection bias when randomization is imperfect. We are also actively interested in research on adaptive experimentation such as Bayesian optimization and reinforcement learning.
- Anomaly Detection. Facebook solicits research on detecting usage pattern on websites or apps that are automated (e.g., scripts and bots). Research should focus on building statistical methods to analyze traffic patterns, and detecting and classifying “regular” user vs. automated behavior patterns in the absence of labeled ground truth data.
- Interpretability techniques for AI models. Complex AI models have become increasingly opaque, so it is important for both the AI practitioners and business stakeholders to have comprehensive evaluation and understanding of the AI algorithms. This can help simplify model development, and make sure we leverage AI responsibly. We are interested in interpretability techniques in (but not limited to) any of the following topics: feature attributions, aggregate attributions, feature interaction, accumulated local effects, and global/local surrogate.
Facebook is pleased to invite faculty to respond to this call for research proposals. In order to support academic work that addresses our challenges and opportunities while producing generalizable knowledge, Facebook is pleased to offer six research awards of up to $50K each. Awards will be made as unrestricted gifts to the principal investigator’s host university.
Proposals should include
- A summary of the project (1-2 pages): Include an explanation of the area of focus, a description of techniques, any relevant prior work, and a timeline with milestones and expected outcomes. Please also include a clear and concise statement of the scientific contribution and routes to eventual deployment.
- A draft budget description (1 page) Include an approximate cost of the project and explanation of how funds would be spent.
- Curriculum Vitae: Provide the name of each researcher involved in the proposed work with their CV/résumé or link to Google Scholar page.
- Administrative information: Include the institution’s tax ID number and details for both a finance contact and authorized authority that can sign award documentation.
- Keyword: Please tag your submission with 1-2 keywords to help us route your proposal to the appropriate reviewers.
- Awards must comply with applicable US and international laws, regulations and policies.
- Applicants must be current faculty at an accredited academic institution or researchers at a research institute or non-profit organization.
- Applicants must be the Principal Investigator on any resulting award.
Timing and dates
- Applications are now open. Deadline to apply is March 25, 2020 at 5:00 p.m. AOE.
- Notifications will be sent by email to selected applicants in June 2020.