BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CMSA - ECPv6.15.18//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:CMSA
X-ORIGINAL-URL:https://cmsa.fas.harvard.edu
X-WR-CALDESC:Events for CMSA
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20210314T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20211107T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20220313T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20221106T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20230312T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20231105T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220521T090000
DTEND;TZID=America/New_York:20220612T170000
DTSTAMP:20260409T125528
CREATED:20230706T182609Z
LAST-MODIFIED:20240229T094452Z
UID:10000147-1653123600-1655053200@cmsa.fas.harvard.edu
SUMMARY:2022 Summer Introduction to Mathematical Research
DESCRIPTION:The Math Department and Harvard’s Center of Mathematical Sciences and Applications (CMSA) will be running a math program/course for mathematically minded undergraduates this summer. The course will be run by Dr. Yingying Wu from CMSA. Here is a description: \nSummer Introduction to Mathematical Research (sponsored by CMSA and the Harvard Math Department) \nIn this course\, we will start with an introduction to computer programming\, algorithms\, and scientific computing. Then we will discuss topics in topology\, classical geometry\, projective geometry\, and differential geometry\, and see how they can be applied to machine learning. We will go on to discuss fundamental concepts of deep learning\, different deep neural network models\, and mathematical interpretations of why deep neural networks are effective from a calculus viewpoint. We will conclude the course with a gentle introduction to cryptography\, introducing some of the iconic topics: Yao’s Millionaires’ problem\, zero-knowledge proof\, the multi-party computation algorithm\, and its proof. \nThe program hopes to provide several research mentors from various disciplines who will give some of the course lectures. Students will have the opportunity to work with one of the research mentors offered by the program. \nPrerequisites: Basic coding ability in some programming language (C/Python/Matlab or CS50 experience). Some background in calculus and linear algebra is needed too. If you wish to work with a research mentor on differential geometry\, more background in geometry such as from Math 132 or 136 will be useful. If you wish to work with a research mentor on computer science\, coding experience mentioned above will be very useful. If you wish to work with a medical scientist\, some background in life science or basic organic chemistry is recommended. \nThe course will meet 3 hours per week for 7 weeks via Zoom on days and times that will be scheduled for the convenience of the participants. There may be other times to be arranged for special events. \nThis program is only open to current Harvard undergraduates; both Mathematics concentrators and non-math concentrators are invited to apply. People already enrolled in a Math Department summer tutorial are welcome to partake in this program also. As with the summer tutorials\, there is no association with the Harvard Summer School; and neither Math concentration credit nor Harvard College credit will be given for completing this course. This course has no official Harvard status and enrollment does not qualify you for any Harvard-related perks (such as a place to live if you are in Boston over the summer.) \nHowever: As with the summer tutorials\, those enrolled are eligible* to receive a stipend of $700\, and if you are a Mathematics concentrator\, any written paper for the course can be submitted to fulfill the Math Concentration third-year paper requirement. (*The stipend is not available for people already receiving a stipend via the Math Department’s summer tutorial program\, nor is it available for PRISE participants or participants in the Herchel Smith program.) \nIf you wish to join this program\, please email Cliff Taubes (chtaubes@math.harvard.edu). The enrollment is limited\, so don’t wait too long to apply.
URL:https://cmsa.fas.harvard.edu/event/2022-summer-introduction-to-mathematical-research/
LOCATION:Virtual
CATEGORIES:Event,Programs
ATTACH;FMTTYPE=image/jpeg:https://cmsa.fas.harvard.edu/media/CMSA-2-600x338-1-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220602T161300
DTEND;TZID=America/New_York:20220602T171300
DTSTAMP:20260409T125528
CREATED:20240214T090758Z
LAST-MODIFIED:20240301T102323Z
UID:10002608-1654186380-1654189980@cmsa.fas.harvard.edu
SUMMARY:Fast Point Transformer
DESCRIPTION:Abstract: The recent success of neural networks enables a better interpretation of 3D point clouds\, but processing a large-scale 3D scene remains a challenging problem. Most current approaches divide a large-scale scene into small regions and combine the local predictions together. However\, this scheme inevitably involves additional stages for pre- and post-processing and may also degrade the final output due to predictions in a local perspective. This talk introduces Fast Point Transformer that consists of a new lightweight self-attention layer. Our approach encodes continuous 3D coordinates\, and the voxel hashing-based architecture boosts computational efficiency. The proposed method is demonstrated with 3D semantic segmentation and 3D detection. The accuracy of our approach is competitive to the best voxel-based method\, and our network achieves 129 times faster inference time than the state-of-the-art\, Point Transformer\, with a reasonable accuracy trade-off in 3D semantic segmentation on S3DIS dataset. \nBio: Jaesik Park is an Assistant Professor at POSTECH. He received his Bachelor’s degree from Hanyang University in 2009\, and he received his Master’s degree and Ph.D. degree from KAIST in 2011 and 2015\, respectively. Before joining POSTECH\, He worked at Intel as a research scientist\, where he co-created the Open3D library. His research interests include image synthesis\, scene understanding\, and 3D reconstruction. He serves as a program committee at prestigious computer vision conferences\, such as Area Chair for ICCV\, CVPR\, and ECCV.
URL:https://cmsa.fas.harvard.edu/event/6-2-2022-interdisciplinary-science-seminar/
LOCATION:MA
CATEGORIES:Interdisciplinary Science Seminar
ATTACH;FMTTYPE=image/jpeg:https://cmsa.fas.harvard.edu/media/CMSA-Interdisciplinary-Science-Seminar-06.02.2022-1583x2048-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220606T090000
DTEND;TZID=America/New_York:20220608T170000
DTSTAMP:20260409T125528
CREATED:20230706T182850Z
LAST-MODIFIED:20250305T172950Z
UID:10000893-1654506000-1654707600@cmsa.fas.harvard.edu
SUMMARY:Symposium on Foundations of Responsible Computing (FORC)
DESCRIPTION:On June 6-8\, 2022\, the CMSA hosted the 3rd annual Symposium on Foundations of Responsible Computing (FORC). \nThe Symposium on Foundations of Responsible Computing (FORC) is a forum for mathematical research in computation and society writ large.  The Symposium aims to catalyze the formation of a community supportive of the application of theoretical computer science\, statistics\, economics and other relevant analytical fields to problems of pressing and anticipated societal concern. \nOrganizers: Cynthia Dwork\, Harvard SEAS | Omer Reingold\, Stanford | Elisa Celis\, Yale \nSchedule\nJune 6\, 2022 \n\n\n\n\n9:15 am–10:15 am\nOpening Remarks \nKeynote Speaker: Caroline Nobo\, Yale University\nTitle: From Theory to Impact: Why Better Data Systems are Necessary for Criminal Legal Reform \nAbstract: This talk will dive into the messy\, archaic\, and siloed world of local criminal justice data in America. We will start with a 30\,000 foot discussion about the current state of criminal legal data systems\, then transition to the challenges of this broken paradigm\, and conclude with a call to measure new things – and to measure them better! This talk will leave you with an understanding of criminal justice data infrastructure and transparency in the US\, and will discuss how expensive case management software and other technology are built on outdated normative values which impede efforts to reform the system. The result is an infuriating paradox: an abundance of tech products built without theoretical grounding\, in a space rich with research and evidence.\n\n\n10:15 am–10:45 am\nCoffee Break\n\n\n\n10:45 am–12:15 pm\nPaper Session 1\nSession Chair: Ruth Urner\n\n\n\nGeorgy Noarov\, University of Pennsylvania\nTitle: Online Minimax Multiobjective Optimization \nAbstract: We introduce a simple but general online learning framework in which a learner plays against an adversary in a vector-valued game that changes every round. The learner’s objective is to minimize the maximum cumulative loss over all coordinates. We give a simple algorithm that lets the learner do almost as well as if she knew the adversary’s actions in advance. We demonstrate the power of our framework by using it to (re)derive optimal bounds and efficient algorithms across a variety of domains\, ranging from multicalibration to a large set of no-regret algorithms\, to a variant of Blackwell’s approachability theorem for polytopes with fast convergence rates. As a new application\, we show how to “(multi)calibeat” an arbitrary collection of forecasters — achieving an exponentially improved dependence on the number of models we are competing against\, compared to prior work.\n\n\n\nMatthew Eichhorn\, Cornell University\nTitle: Mind your Ps and Qs: Allocation with Priorities and Quotas \nAbstract: In many settings\, such as university admissions\, the rationing of medical supplies\, and the assignment of public housing\, decision-makers use normative criteria (ethical\, financial\, legal\, etc.) to justify who gets an allocation. These criteria can often be translated into quotas for the number of units available to particular demographics and priorities over agents who qualify in each demographic. Each agent may qualify in multiple categories at different priority levels\, so many allocations may conform to a given set of quotas and priorities. Which of these allocations should be chosen? In this talk\, I’ll formalize this reserve allocation problem and motivate Pareto efficiency as a natural desideratum. I’ll present an algorithm to locate efficient allocations that conform to the quota and priority constraints. This algorithm relies on beautiful techniques from integer and linear programming\, and it is both faster and more straightforward than existing techniques in this space. Moreover\, its clean formulation allows for further refinement\, such as the secondary optimization of some heuristics for fairness.\n\n\n\nHaewon Jeong\, Harvard University\nTitle: Fairness without Imputation: A Decision Tree Approach for Fair Prediction with Missing Values \nAbstract: We investigate the fairness concerns of training a machine learning model using data with missing values. Even though there are a number of fairness intervention methods in the literature\, most of them require a complete training set as input. In practice\, data can have missing values\, and data missing patterns can depend on group attributes (e.g. gender or race). Simply applying off-the-shelf fair learning algorithms to an imputed dataset may lead to an unfair model. In this paper\, we first theoretically analyze different sources of discrimination risks when training with an imputed dataset. Then\, we propose an integrated approach based on decision trees that does not require a separate process of imputation and learning. Instead\, we train a tree with missing incorporated as attribute (MIA)\, which does not require explicit imputation\, and we optimize a fairness-regularized objective function. We demonstrate that our approach outperforms existing fairness intervention methods applied to an imputed dataset\, through several experiments on real-world datasets.\n\n\n\nEmily Diana\, University of Pennsylvania\nTitle: Multiaccurate Proxies for Downstream Fairness \nAbstract: We study the problem of training a model that must obey demographic fairness conditions when the sensitive features are not available at training time — in other words\, how can we train a model to be fair by race when we don’t have data about race? We adopt a fairness pipeline perspective\, in which an “upstream” learner that does have access to the sensitive features will learn a proxy model for these features from the other attributes. The goal of the proxy is to allow a general “downstream” learner — with minimal assumptions on their prediction task — to be able to use the proxy to train a model that is fair with respect to the true sensitive features. We show that obeying multiaccuracy constraints with respect to the downstream model class suffices for this purpose\, provide sample- and oracle efficient-algorithms and generalization bounds for learning such proxies\, and conduct an experimental evaluation. In general\, multiaccuracy is much easier to satisfy than classification accuracy\, and can be satisfied even when the sensitive features are hard to predict.\n\n\n12:15 pm–1:45 pm\nLunch Break\n\n\n\n1:45–3:15 pm\nPaper Session 2\nSession Chair: Guy Rothblum\n\n\n\nElbert Du\, Harvard University\nTitle: Improved Generalization Guarantees in Restricted Data Models \nAbstract: Differential privacy is known to protect against threats to validity incurred due to adaptive\, or exploratory\, data analysis — even when the analyst adversarially searches for a statistical estimate that diverges from the true value of the quantity of interest on the underlying population. The cost of this protection is the accuracy loss incurred by differential privacy. In this work\, inspired by standard models in the genomics literature\, we consider data models in which individuals are represented by a sequence of attributes with the property that where distant attributes are only weakly correlated. We show that\, under this assumption\, it is possible to “re-use” privacy budget on different portions of the data\, significantly improving accuracy without increasing the risk of overfitting.\n\n\n\nRuth Urner\, York University\nTitle: Robustness Should not be at Odds with Accuracy \nAbstract: The phenomenon of adversarial examples in deep learning models has caused substantial concern over their reliability and trustworthiness: in many instances an imperceptible perturbation can falsely flip a neural network’s prediction. Applied research in this area has mostly focused on developing novel adversarial attack strategies or building better defenses against such. It has repeatedly been pointed out that adversarial robustness may be in conflict with requirements for high accuracy. In this work\, we take a more principled look at modeling the phenomenon of adversarial examples. We argue that deciding whether a model’s label change under a small perturbation is justified\, should be done in compliance with the underlying data-generating process. Through a series of formal constructions\, systematically analyzing the the relation between standard Bayes classifiers and robust-Bayes classifiers\, we make the case for adversarial robustness as a locally adaptive measure. We propose a novel way defining such a locally adaptive robust loss\, show that it has a natural empirical counterpart\, and develop resulting algorithmic guidance in form of data-informed adaptive robustness radius. We prove that our adaptive robust data-augmentation maintains consistency of 1-nearest neighbor classification under deterministic labels and thereby argue that robustness should not be at odds with accuracy.\n\n\n\nSushant Agarwal\, University of Waterloo\nTitle: Towards the Unification and Robustness of Perturbation and Gradient Based Explanations \nAbstract: As machine learning black boxes are increasingly being deployed in critical domains such as healthcare and criminal justice\, there has been a growing emphasis on developing techniques for explaining these black boxes in a post hoc manner. In this work\, we analyze two popular post hoc interpretation techniques: SmoothGrad which is a gradient based method\, and a variant of LIME which is a perturbation based method. More specifically\, we derive explicit closed form expressions for the explanations output by these two methods and show that they both converge to the same explanation in expectation\, i.e.\, when the number of perturbed samples used by these methods is large. We then leverage this connection to establish other desirable properties\, such as robustness and linearity\, for these techniques. We also derive finite sample complexity bounds for the number of perturbations required for these methods to converge to their expected explanation. Finally\, we empirically validate our theory using extensive experimentation on both synthetic and real world datasets.\n\n\n\nTijana Zrnic\, University of California\, Berkeley\nTitle: Regret Minimization with Performative Feedback \nAbstract: In performative prediction\, the deployment of a predictive model triggers a shift in the data distribution. As these shifts are typically unknown ahead of time\, the learner needs to deploy a model to get feedback about the distribution it induces. We study the problem of finding near-optimal models under performativity while maintaining low regret. On the surface\, this problem might seem equivalent to a bandit problem. However\, it exhibits a fundamentally richer feedback structure that we refer to as performative feedback: after every deployment\, the learner receives samples from the shifted distribution rather than only bandit feedback about the reward. Our main contribution is regret bounds that scale only with the complexity of the distribution shifts and not that of the reward function. The key algorithmic idea is careful exploration of the distribution shifts that informs a novel construction of confidence bounds on the risk of unexplored models. The construction only relies on smoothness of the shifts and does not assume convexity. More broadly\, our work establishes a conceptual approach for leveraging tools from the bandits literature for the purpose of regret minimization with performative feedback.\n\n\n3:15 pm–3:45 pm\nCoffee Break\n\n\n\n3:45 pm–5:00 pm\nPanel Discussion\nTitle: What is Responsible Computing? \nPanelists: Jiahao Chen\, Cynthia Dwork\, Kobbi Nissim\, Ruth Urner \nModerator: Elisa Celis\n\n\n\n\n  \nJune 7\, 2022 \n\n\n\n\n9:15 am–10:15 am\nKeynote Speaker: Isaac Kohane\, Harvard Medical School\nTitle: What’s in a label? The case for and against monolithic group/ethnic/race labeling for machine learning \nAbstract: Populations and group labels have been used and abused for thousands of years. The scale at which AI can incorporate such labels into its models and the ways in which such models can be misused are cause for significant concern. I will describe\, with examples drawn from experiments in precision medicine\, the task dependence of how underserved and oppressed populations can be both harmed and helped by the use of group labels. The source of the labels and the utility models underlying their use will be particularly emphasized.\n\n\n10:15 am–10:45 am\nCoffee Break\n\n\n\n10:45 am–12:15 pm\nPaper Session 3\nSession Chair: Ruth Urner\n\n\n\nRojin Rezvan\, University of Texas at Austin\nTitle: Individually-Fair Auctions for Multi-Slot Sponsored Search \nAbstract: We design fair-sponsored search auctions that achieve a near-optimal tradeoff between fairness and quality. Our work builds upon the model and auction design of Chawla and Jagadeesan\, who considered the special case of a single slot. We consider sponsored search settings with multiple slots and the standard model of click-through rates that are multiplicatively separable into an advertiser-specific component and a slot-specific component. When similar users have similar advertiser-specific click-through rates\, our auctions achieve the same near-optimal tradeoff between fairness and quality. When similar users can have different advertiser-specific preferences\, we show that a preference-based fairness guarantee holds. Finally\, we provide a computationally efficient algorithm for computing payments for our auctions as well as those in previous work\, resolving another open direction from Chawla and Jagadeesan.\n\n\n\nJudy Hanwen Shen\, Stanford\nTitle: Leximax Approximations and Representative Cohort Selection \nAbstract: Finding a representative cohort from a broad pool of candidates is a goal that arises in many contexts such as choosing governing committees and consumer panels. While there are many ways to define the degree to which a cohort represents a population\, a very appealing solution concept is lexicographic maximality (leximax) which offers a natural (pareto-optimal like) interpretation that the utility of no population can be increased without decreasing the utility of a population that is already worse off. However\, finding a leximax solution can be highly dependent on small variations in the utility of certain groups. In this work\, we explore new notions of approximate leximax solutions with three distinct motivations: better algorithmic efficiency\, exploiting significant utility improvements\, and robustness to noise. Among other definitional contributions\, we give a new notion of an approximate leximax that satisfies a similarly appealing semantic interpretation and relate it to algorithmically-feasible approximate leximax notions. When group utilities are linear over cohort candidates\, we give an efficient polynomial-time algorithm for finding a leximax distribution over cohort candidates in the exact as well as in the approximate setting. Furthermore\, we show that finding an integer solution to leximax cohort selection with linear utilities is NP-Hard.\n\n\n\nJiayuan Ye\,\nNational University of Singapore\nTitle: Differentially Private Learning Needs Hidden State (or Much Faster Convergence) \nAbstract: Differential privacy analysis of randomized learning algorithms typically relies on composition theorems\, where the implicit assumption is that the internal state of the iterative algorithm is revealed to the adversary. However\, by assuming hidden states for DP algorithms (when only the last-iterate is observable)\, recent works prove a converging privacy bound for noisy gradient descent (on strongly convex smooth loss function) that is significantly smaller than composition bounds after a few epochs. In this talk\, we extend this hidden-state analysis to various stochastic minibatch gradient descent schemes (such as under “shuffle and partition” and “sample without replacement”)\, by deriving novel bounds for the privacy amplification by random post-processing and subsampling. We prove that\, in these settings\, our privacy bound is much smaller than composition for training with a large number of iterations (which is the case for learning from high-dimensional data). Our converging privacy analysis\, thus\, shows that differentially private learning\, with a tight bound\, needs hidden state privacy analysis or a fast convergence. To complement our theoretical results\, we present experiments for training classification models on MNIST\, FMNIST and CIFAR-10 datasets\, and observe a better accuracy given fixed privacy budgets\, under the hidden-state analysis.\n\n\n\nMahbod Majid\, University of Waterloo\nTitle: Efficient Mean Estimation with Pure Differential Privacy via a Sum-of-Squares Exponential Mechanism \nAbstract: We give the first polynomial-time algorithm to estimate the mean of a d-variate probability distribution from O(d) independent samples (up to logarithmic factors) subject to pure differential privacy. \nOur main technique is a new approach to use the powerful Sum of Squares method (SoS) to design differentially private algorithms. SoS proofs to algorithms is a key theme in numerous recent works in high-dimensional algorithmic statistics – estimators which apparently require exponential running time but whose analysis can be captured by low-degree Sum of Squares proofs can be automatically turned into polynomial-time algorithms with the same provable guarantees. We demonstrate a similar proofs to private algorithms phenomenon: instances of the workhorse exponential mechanism which apparently require exponential time but which can be analyzed with low-degree SoS proofs can be automatically turned into polynomial-time differentially private algorithms. We prove a meta-theorem capturing this phenomenon\, which we expect to be of broad use in private algorithm design.\n\n\n12:15 pm–1:45 pm\nLunch Break\n\n\n\n1:45–3:15 pm\nPaper Session 4\nSession Chair: Kunal Talwar\n\n\n\nKunal Talwar\,\nApple\nTitle: Differential Secrecy for Distributed Data and Applications to Robust Differentially Secure Vector Summation \nAbstract: Computing the noisy sum of real-valued vectors is an important primitive in differentially private learning and statistics. In private federated learning applications\, these vectors are held by client devices\, leading to a distributed summation problem. Standard Secure Multiparty Computation (SMC) protocols for this problem are susceptible to poisoning attacks\, where a client may have a large influence on the sum\, without being detected.\nIn this work\, we propose a poisoning-robust private summation protocol in the multiple-server setting\, recently studied in PRIO. We present a protocol for vector summation that verifies that the Euclidean norm of each contribution is approximately bounded. We show that by relaxing the security constraint in SMC to a differential privacy like guarantee\, one can improve over PRIO in terms of communication requirements as well as the client-side computation. Unlike SMC algorithms that inevitably cast integers to elements of a large finite field\, our algorithms work over integers/reals\, which may allow for additional efficiencies.\n\n\n\nGiuseppe Vietri\, University of Minnesota\nTitle: Improved Regret for Differentially Private Exploration in Linear MDP \nAbstract: We study privacy-preserving exploration in sequential decision-making for environments that rely on sensitive data such as medical records. In particular\, we focus on solving the problem of reinforcement learning (RL) subject to the constraint of (joint) differential privacy in the linear MDP setting\, where both dynamics and rewards are given by linear functions. Prior work on this problem due to Luyo et al. (2021) achieves a regret rate that has a dependence of O(K^{3/5}) on the number of episodes K. We provide a private algorithm with an improved regret rate with an optimal dependence of O(K^{1/2}) on the number of episodes. The key recipe for our stronger regret guarantee is the adaptivity in the policy update schedule\, in which an update only occurs when sufficient changes in the data are detected. As a result\, our algorithm benefits from low switching cost and only performs O(log(K)) updates\, which greatly reduces the amount of privacy noise. Finally\, in the most prevalent privacy regimes where the privacy parameter ? is a constant\, our algorithm incurs negligible privacy cost — in comparison with the existing non-private regret bounds\, the additional regret due to privacy appears in lower-order terms.\n\n\n\nMingxun Zhou\,\nCarnegie Mellon University\nTitle: The Power of the Differentially Oblivious Shuffle in Distributed Privacy MechanismsAbstract: The shuffle model has been extensively investigated in the distributed differential privacy (DP) literature. For a class of useful computational tasks\, the shuffle model allows us to achieve privacy-utility tradeoff similar to those in the central model\, while shifting the trust from a central data curator to a “trusted shuffle” which can be implemented through either trusted hardware or cryptography. Very recently\, several works explored cryptographic instantiations of a new type of shuffle with relaxed security\, called differentially oblivious (DO) shuffles. These works demonstrate that by relaxing the shuffler’s security from simulation-style secrecy to differential privacy\, we can achieve asymptotical efficiency improvements. A natural question arises\, can we replace the shuffler in distributed DP mechanisms with a DO-shuffle while retaining a similar privacy-utility tradeoff?\nIn this paper\, we prove an optimal privacy amplification theorem by composing any locally differentially private (LDP) mechanism with a DO-shuffler\, achieving parameters that tightly match the shuffle model. Moreover\, we explore multi-message protocols in the DO-shuffle model\, and construct mechanisms for the real summation and histograph problems. Our error bounds approximate the best known results in the multi-message shuffle-model up to sub-logarithmic factors. Our results also suggest that just like in the shuffle model\, allowing each client to send multiple messages is fundamentally more powerful than restricting to a single message.\n\n\n\nBadih Ghazi\,\nGoogle Research\nTitle: Differentially Private Ad Conversion Measurement \nAbstract: In this work\, we study conversion measurement\, a central functionality in the digital advertising space\, where an advertiser seeks to estimate advertiser site conversions attributed to ad impressions that users have interacted with on various publisher sites. We consider differential privacy (DP)\, a notion that has gained in popularity due to its strong and rigorous guarantees\, and suggest a formal framework for DP conversion measurement\, uncovering a subtle interplay between attribution and privacy. We define the notion of an operationally valid configuration of the attribution logic\, DP adjacency relation\, privacy\nbudget scope and enforcement point\, and provide\, for a natural space of configurations\, a complete characterization.\n\n\n3:15 pm–3:45 pm\nCoffee Break\n\n\n\n3:45 pm–5:00 pm\nOpen Poster Session\n\n\n\n\n\n  \nJune 8\, 2022 \n\n\n\n\n9:15 am–10:15 am\nKeynote Speaker: Nuria Oliver\, Data-Pop Alliance\nTitle: Data Science against COVID-19 \nAbstract: In my talk\, I will describe the work that I have been doing since March 2020\, leading a multi-disciplinary team of 20+ volunteer scientists working very closely with the Presidency of the Valencian Government in Spain on 4 large areas: (1) human mobility modeling; (2) computational epidemiological models (both metapopulation\, individual and LSTM-based models); (3) predictive models; and (4) citizen surveys via the COVID19impactsurvey with over 600\,000 answers worldwide. \nI will describe the results that we have produced in each of these areas\, including winning the 500K XPRIZE Pandemic Response Challenge and best paper award at ECML-PKDD 2021. I will share the lessons learned in this very special initiative of collaboration between the civil society at large (through the survey)\, the scientific community (through the Expert Group) and a public administration (through the Commissioner at the Presidency level). WIRED magazine just published an article describing our story.\n\n\n10:15 am–10:45 am\nCoffee Break\n\n\n\n10:45 am–12:15 pm\nPaper Session 5\nSession Chair: Kunal Talwar\n\n\n\nShengyuan Hu\, Carnegie Mellon University\nTitle: Private Multi-Task Learning: Formulation and Applications to Federated Learning \nAbstract: Many problems in machine learning rely on multi-task learning (MTL)\, in which the goal is to solve multiple related machine learning tasks simultaneously. MTL is particularly relevant for privacy-sensitive applications in areas such as healthcare\, finance\, and IoT computing\, where sensitive data from multiple\, varied sources are shared for the purpose of learning. In this work\, we formalize notions of task-level privacy for MTL via joint differential privacy (JDP)\, a relaxation of differential privacy for mechanism design and distributed optimization. We then propose an algorithm for mean-regularized MTL\, an objective commonly used for applications in personalized federated learning\, subject to JDP. We analyze our objective and solver\, providing certifiable guarantees on both privacy and utility. Empirically\, our method allows for improved privacy/utility trade-offs relative to global baselines across common federated learning benchmarks\n\n\n\nChristina Yu\,\nCornell University\nTitle: Sequential Fair Allocation: Achieving the Optimal Envy-Efficiency Tradeoff Curve \nAbstract: We consider the problem of dividing limited resources to individuals arriving over T rounds with a goal of achieving fairness across individuals. In general there may be multiple resources and multiple types of individuals with different utilities. A standard definition of `fairness’ requires an allocation to simultaneously satisfy envy-freeness and Pareto efficiency. However\, in the online sequential setting\, the social planner must decide on a current allocation before the downstream demand is realized\, such that no policy can guarantee these desiderata simultaneously with probability 1\, requiring a modified metric of measuring fairness for online policies. We show that in the online setting\, the two desired properties (envy-freeness and efficiency) are in direct contention\, in that any algorithm achieving additive counterfactual envy-freeness up to L_T necessarily suffers an efficiency loss of at least 1 / L_T. We complement this uncertainty principle with a simple algorithm\, HopeGuardrail\, which allocates resources based on an adaptive threshold policy and is able to achieve any fairness-efficiency point on this frontier. Our result is the first to provide guarantees for fair online resource allocation with high probability for multiple resource and multiple type settings. In simulation results\, our algorithm provides allocations close to the optimal fair solution in hindsight\, motivating its use in practical applications as the algorithm is able to adapt to any desired fairness efficiency trade-off.\n\n\n\nHedyeh Beyhaghi\, Carnegie Mellon University\nTitle: On classification of strategic agents who can both game and improve \nAbstract: In this work\, we consider classification of agents who can both game and improve. For example\, people wishing to get a loan may be able to take some actions that increase their perceived credit-worthiness and others that also increase their true credit-worthiness. A decision-maker would like to define a classification rule with few false-positives (does not give out many bad loans) while yielding many true positives (giving out many good loans)\, which includes encouraging agents to improve to become true positives if possible. We consider two models for this problem\, a general discrete model and a linear model\, and prove algorithmic\, learning\, and hardness results for each. For the general discrete model\, we give an efficient algorithm for the problem of maximizing the number of true positives subject to no false positives\, and show how to extend this to a partial-information learning setting. We also show hardness for the problem of maximizing the number of true positives subject to a nonzero bound on the number of false positives\, and that this hardness holds even for a finite-point version of our linear model. We also show that maximizing the number of true positives subject to no false positive is NP-hard in our full linear model. We additionally provide an algorithm that determines whether there exists a linear classifier that classifies all agents accurately and causes all improvable agents to become qualified\, and give additional results for low-dimensional data.\n\n\n\nKeegan Harris\, Carnegie Mellon University\nTitle: Bayesian Persuasion for Algorithmic Recourse \nAbstract: When subjected to automated decision-making\, decision subjects may strategically modify their observable features in ways they believe will maximize their chances of receiving a favorable decision. In many practical situations\, the underlying assessment rule is deliberately kept secret to avoid gaming and maintain competitive advantage. The resulting opacity forces the decision subjects to rely on incomplete information when making strategic feature modifications. We capture such settings as a game of Bayesian persuasion\, in which the decision maker offers a form of recourse to the decision subject by providing them with an action recommendation (or signal) to incentivize them to modify their features in desirable ways. We show that when using persuasion\, both the decision maker and decision subject are never worse off in expectation\, while the decision maker can be significantly better off. While the decision maker’s problem of finding the optimal Bayesian incentive-compatible (BIC) signaling policy takes the form of optimization over infinitely-many variables\, we show that this optimization can be cast as a linear program over finitely-many regions of the space of possible assessment rules. While this reformulation simplifies the problem dramatically\, solving the linear program requires reasoning about exponentially-many variables\, even under relatively simple settings. Motivated by this observation\, we provide a polynomial-time approximation scheme that recovers a near-optimal signaling policy. Finally\, our numerical simulations on semi-synthetic data empirically illustrate the benefits of using persuasion in the algorithmic recourse setting.\n\n\n12:15 pm–1:45 pm\nLunch Break\n\n\n\n1:45–3:15 pm\nPaper Session 6\nSession Chair: Elisa Celis\n\n\n\nMark Bun\, Boston University\nTitle: Controlling Privacy Loss in Sampling Schemes: An Analysis of Stratified and Cluster Sampling \nAbstract: Sampling schemes are fundamental tools in statistics\, survey design\, and algorithm design. A fundamental result in differential privacy is that a differentially private mechanism run on a simple random sample of a population provides stronger privacy guarantees than the same algorithm run on the entire population. However\, in practice\, sampling designs are often more complex than the simple\, data-independent sampling schemes that are addressed in prior work. In this work\, we extend the study of privacy amplification results to more complex\, data-dependent sampling schemes. We find that not only do these sampling schemes often fail to amplify privacy\, they can actually result in privacy degradation. We analyze the privacy implications of the pervasive cluster sampling and stratified sampling paradigms\, as well as provide some insight into the study of more general sampling designs.\n\n\n\nSamson Zhou\, Carnegie Mellon University\nTitle: Private Data Stream Analysis for Universal Symmetric Norm Estimation \nAbstract: We study how to release summary statistics on a data stream subject to the constraint of differential privacy. In particular\, we focus on releasing the family of symmetric norms\, which are invariant under sign-flips and coordinate-wise permutations on an input data stream and include L_p norms\, k-support norms\, top-k norms\, and the box norm as special cases. Although it may be possible to design and analyze a separate mechanism for each symmetric norm\, we propose a general parametrizable framework that differentially privately releases a number of sufficient statistics from which the approximation of all symmetric norms can be simultaneously computed. Our framework partitions the coordinates of the underlying frequency vector into different levels based on their magnitude and releases approximate frequencies for the “heavy” coordinates in important levels and releases approximate level sizes for the “light” coordinates in important levels. Surprisingly\, our mechanism allows for the release of an arbitrary number of symmetric norm approximations without any overhead or additional loss in privacy. Moreover\, our mechanism permits (1+\alpha)-approximation to each of the symmetric norms and can be implemented using sublinear space in the streaming model for many regimes of the accuracy and privacy parameters.\n\n\n\nAloni Cohen\, University of Chicago\nTitle: Attacks on Deidentification’s Defenses \nAbstract: Quasi-identifier-based deidentification techniques (QI-deidentification) are widely used in practice\, including k-anonymity\, ?-diversity\, and t-closeness. We present three new attacks on QI-deidentification: two theoretical attacks and one practical attack on a real dataset. In contrast to prior work\, our theoretical attacks work even if every attribute is a quasi-identifier. Hence\, they apply to k-anonymity\, ?-diversity\, t-closeness\, and most other QI-deidentification techniques.\nFirst\, we introduce a new class of privacy attacks called downcoding attacks\, and prove that every QI-deidentification scheme is vulnerable to downcoding attacks if it is minimal and hierarchical. Second\, we convert the downcoding attacks into powerful predicate singling-out (PSO) attacks\, which were recently proposed as a way to demonstrate that a privacy mechanism fails to legally anonymize under Europe’s General Data Protection Regulation. Third\, we use LinkedIn.com to reidentify 3 students in a k-anonymized dataset published by EdX (and show thousands are potentially vulnerable)\, undermining EdX’s claimed compliance with the Family Educational Rights and Privacy Act. \nThe significance of this work is both scientific and political. Our theoretical attacks demonstrate that QI-deidentification may offer no protection even if every attribute is treated as a quasi-identifier. Our practical attack demonstrates that even deidentification experts acting in accordance with strict privacy regulations fail to prevent real-world reidentification. Together\, they rebut a foundational tenet of QI-deidentification and challenge the actual arguments made to justify the continued use of k-anonymity and other QI-deidentification techniques.\n\n\n\nSteven Wu\,\nCarnegie Mellon University\nTitle: Fully Adaptive Composition in Differential Privacy \nAbstract: Composition is a key feature of differential privacy. Well-known advanced composition theorems allow one to query a private database quadratically more times than basic privacy composition would permit. However\, these results require that the privacy parameters of all algorithms be fixed before interacting with the data. To address this\, Rogers et al. introduced fully adaptive composition\, wherein both algorithms and their privacy parameters can be selected adaptively. The authors introduce two probabilistic objects to measure privacy in adaptive composition: privacy filters\, which provide differential privacy guarantees for composed interactions\, and privacy odometers\, time-uniform bounds on privacy loss. There are substantial gaps between advanced composition and existing filters and odometers. First\, existing filters place stronger assumptions on the algorithms being composed. Second\, these odometers and filters suffer from large constants\, making them impractical. We construct filters that match the tightness of advanced composition\, including constants\, despite allowing for adaptively chosen privacy parameters. We also construct several general families of odometers. These odometers can match the tightness of advanced composition at an arbitrary\, preselected point in time\, or at all points in time simultaneously\, up to a doubly-logarithmic factor. We obtain our results by leveraging recent advances in time-uniform martingale concentration. In sum\, we show that fully adaptive privacy is obtainable at almost no loss\, and conjecture that our results are essentially not improvable (even in constants) in general.\n\n\n3:15 pm–3:45 pm\nFORC Reception\n\n\n\n3:45 pm–5:00 pm\nSocial Hour
URL:https://cmsa.fas.harvard.edu/event/symposium-on-foundations-of-responsible-computing-forc/
LOCATION:CMSA Room G10\, CMSA\, 20 Garden Street\, Cambridge\, MA\, 02138\, United States
CATEGORIES:Conference,Event
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/FORC22_poster.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220616T090000
DTEND;TZID=America/New_York:20220616T100000
DTSTAMP:20260409T125528
CREATED:20240215T094047Z
LAST-MODIFIED:20240229T084329Z
UID:10002724-1655370000-1655373600@cmsa.fas.harvard.edu
SUMMARY:Surface hopping algorithms for non-adiabatic quantum systems
DESCRIPTION:Interdisciplinary Science Seminar\n\n\n\n\nSpeaker: Jianfeng Lu\, Duke UniversityTitle: Surface hopping algorithms for non-adiabatic quantum systems \nAbstract: Surface hopping algorithm is widely used in chemistry for mixed quantum-classical dynamics. In this talk\, we will discuss some of our recent works in mathematical understanding and algorithm development for surface hopping methods. These methods are based on stochastic approximations of semiclassical path-integral representation to the solution of multi-level Schrodinger equations; such methodology also extends to other high-dimensional transport systems.
URL:https://cmsa.fas.harvard.edu/event/iss_61622/
LOCATION:CMSA Room G10\, CMSA\, 20 Garden Street\, Cambridge\, MA\, 02138\, United States
CATEGORIES:Interdisciplinary Science Seminar
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220621T090000
DTEND;TZID=America/New_York:20220624T170000
DTSTAMP:20260409T125528
CREATED:20230706T183302Z
LAST-MODIFIED:20250305T175141Z
UID:10000894-1655802000-1656090000@cmsa.fas.harvard.edu
SUMMARY:Joint BHI/CMSA Conference on Flat Holography
DESCRIPTION:On June 21–24\, 2022\, the Harvard Black Hole Initiative and the CMSA hosted the Joint BHI/CMSA Conference on Flat Holography (and related topics). \nThe recent discovery of infinitely-many soft symmetries for all quantum theories of gravity in asymptotically flat space has provided a promising starting point for a bottom-up construction of a holographic dual for the real world. Recent developments have brought together previously disparate studies of soft theorems\, asymptotic symmetries\, twistor theory\, asymptotically flat black holes and their microscopic duals\, self-dual gravity\, and celestial scattering amplitudes\, and link directly to AdS/CFT. \nThe conference was held in room G10 of the CMSA\, 20 Garden Street\, Cambridge\, MA. \nOrganizers: \n\nDaniel Kapec\, CMSA\nAndrew Strominger\, BHI\nShing-Tung Yau\, Harvard & Tsinghua\n\nConfirmed Speakers: \n\nNima Arkani-Hamed\, IAS\nShamik Banerjee\, Bhubaneswar\, Inst. Phys.\nMiguel Campiglia\, Republica U.\, Montevido\nGeoffrey Compere\, Brussels\nLaura Donnay\, Vienna\nNetta Engelhardt\, MIT\nLaurent Freidel\, Perimeter\nAlex Lupsasca\, Princeton\nJuan Maldacena\, IAS\nLionel Mason\, Oxford\nNatalie Paquette\, U. Washington\nSabrina Pasterski\, Princeton/Perimeter\nAndrea Puhm\, Ecole Polytechnique\nAna-Maria Raclariu\, Perimeter\nMarcus Spradlin\, Brown\nTomasz Taylor\, Northeastern\nHerman Verlinde\, Princeton\nAnastasia Volovich\, Brown\nBin Zhu\, Northeastern\n\nShort talks by: Gonçalo Araujo-Regado (Cambridge)\, Adam Ball (Harvard)\, Eduardo Casali (Harvard)\, Jordan Cotler (Harvard)\, Erin Crawley (Harvard)\, Stéphane Detournay (Brussels)\, Alfredo Guevara (Harvard)\, Temple He (UC Davis)\, Elizabeth Himwich (Harvard)\, Yangrui Hu (Brown)\, Daniel Kapec (Harvard)\, Rifath Khan (Cambridge)\, Albert Law (Harvard)\, Luke Lippstreu (Brown)\, Noah Miller (Harvard)\, Sruthi Narayanan (Harvard)\, Lecheng Ren (Brown)\, Francisco Rojas (UAI)\, Romain Ruzziconi (Vienna)\, Andrew Strominger (Harvard)\, Adam Tropper (Harvard)\, Tianli Wang (Harvard)\, Walker Melton (Harvard) \n\n\nSchedule\nMonday\, June 20\, 2022 \n\n\n\n\n\nArrival\n\n\n7:00–9:00 pm\nWelcome Reception at Andy’s residence\n\n\n\n\n  \nTuesday\, June 21\, 2022 \n\n\n\n\n9:00–9:30 am\nBreakfast\nlight breakfast provided\n\n\n\nMorning Session\nChair: Dan Kapec\n\n\n9:30–10:00 am\nHerman Verlinde\nTitle: Comments on Celestial Dynamics\n\n\n10:00–10:30 am\nJuan Maldacena\nTitle: What happens when you spend too much time looking at supersymmetric\nblack holes?\n\n\n10:30–11:00\nCoffee break\n\n\n\n11:00–11:30 am\nMiguel Campiglia\nTitle: Asymptotic symmetries and loop corrections to soft theorems\n\n\n11:30–12:00 pm\nGeoffrey Compere\nTitle: Metric reconstruction from $Lw_{1+\infty}$ multipoles \nAbstract: The most general vacuum solution to Einstein’s field equations with no incoming radiation can be constructed perturbatively from two infinite sets of canonical multipole moments\, which are found to be exchanged under gravitational electric-magnetic duality at the non-linear level. We demonstrate that in non-radiative regions such spacetimes are completely determined by a set of conserved celestial charges\, which uniquely label transitions among non-radiative regions caused by radiative processes. The algebra of the conserved celestial charges is derived from the real $Lw_{1+\infty}$ algebra. The celestial charges are expressed in terms of multipole moments\, which allows to holographically reconstruct the metric in de Donder\, Newman-Unti or Bondi gauge outside of sources.\n\n\n12:00–2:00 pm\nLunch break\n\n\n\n\nAfternoon Session\nChair: Eduardo Casali\n\n\n2:00–2:30 pm\nNatalie Paquette\nTitle: New thoughts on old gauge amplitudes\n\n\n2:30–3:00 pm\nLionel Mason\nTitle: An open sigma model for celestial gravity \nAbstract: A global twistor construction for conformally self-dual split signature metrics on $S2\times S2$  was developed 15 years ago by Claude LeBrun and the speaker.  This encodes the conformal metric into the location of a finite deformation of the real twistor space inside the flat complex twistor space\, $\mathbb{CP}3$. This talk adapts the construction to construct global SD Einstein metrics from conformal boundary data and perturbations around the self-dual sector.  The construction entails determining a family of holomorphic discs in $\mathbb{CP}3$ whose boundaries lie on the deformed real slice and the (chiral) sigma model controls these discs in the Einstein case and provides amplitude formulae.\n\n\n3:00–3:30 pm\nCoffee break\n\n\n\n3:30–4:30 pm\nShort Talks\nDaniel Kapec: Soft Scalars and the Geometry of the Space of Celestial CFTs \nAlbert Law: Soft Scalars and the Geometry of the Space of Celestial CFTs \nSruthi Narayanan: Soft Scalars and the Geometry of the Space of Celestial CFTs \nStéphane Detournay: Non-conformal symmetries and near-extremal black holes \nFrancisco Rojas: Celestial string amplitudes beyond tree level \nTemple He: An effective description of energy transport from holography\n\n\n4:30–5:00 pm\nNima Arkani-Hamed\n(Dual) surfacehedra and flow particles know about strings\n\n\n\n\n  \nWednesday\, June 22\, 2022 \n\n\n\n\n9:00–9:30 am\nBreakfast\nlight breakfast provided\n\n\n\nMorning Session\nChair: Alfredo Guevara\n\n\n9:30–10:00 am\nLaurent Freidel\nTitle: Higher spin symmetry in gravity \nAbstract: In this talk\, I will review how the gravitational conservation laws at infinity reveal a tower of symmetry charges in an asymptotically flat spacetime.\nI will show how the conservation laws\, at spacelike infinity\, give a tower of soft theorems that connect to the ones revealed by celestial holography.\nI’ll present the expression for the symmetry charges in the radiative phase space\, which opens the way to reveal the structure of the algebra beyond the positive helicity sector. Then\, if time permits I’ll browse through many questions that these results raise:\nsuch as the nature of the spacetime symmetry these charges represent\, the nature of the relationship with multipole moments\, and the insights their presence provides for quantum gravity.\n\n\n10:00–10:30 am\nAna-Maria Raclariu\nTitle: Eikonal approximation in celestial CFT\n\n\n10:30–11:00 am\nCoffee break\n\n\n\n11:00–11:30 am\nAnastasia Volovich\nTitle: Effective Field Theories with Celestial Duals\n\n\n11:30–12:00 pm\nMarcus Spradlin\nTitle: Loop level gluon OPE’s in celestial holography\n\n\n12:00–2:00 pm\nLunch break\n\n\n\n\nAfternoon Session\nChair: Chiara Toldo\n\n\n2:00–2:30 pm\nNetta Engelhardt\nTitle: Wormholes from entanglement: true or false?\n\n\n2:30–3:00 pm\nShort Talks\nLuke Lippstreu: Loop corrections to the OPE of celestial gluons \nYangrui Hu: Light transforms of celestial amplitudes \nLecheng Ren: All-order OPE expansion of celestial gluon and graviton primaries from MHV amplitudes\n\n\n3:00–3:30 pm\nCoffee break\n\n\n\n3:30–4:30 pm\nShort Talks\nNoah Miller: C Metric Thermodynamics \nErin Crawley: Kleinian black holes \nRifath Khan: Cauchy Slice Holography: A New AdS/CFT Dictionary \nGonçalo Araujo-Regado: Cauchy Slice Holography: A New AdS/CFT Dictionary \nTianli Wang: Soft Theorem in the BFSS Matrix Model \nAdam Tropper: Soft Theorem in the BFSS Matrix Model\n\n\n7:00–9:00 pm\nBanquet\nMaharaja Restaurant\, 57 JFK Street\, Cambridge\, MA\n\n\n\n\n  \nThursday\, June 23\, 2022 \n\n\n\n\n9:00–9:30 am\nBreakfast\nlight breakfast provided\n\n\n\nMorning Session\nChair: Jordan Cotler\n\n\n9:30–10:00 am\nLaura Donnay\nTitle: A Carrollian road to flat space holography\n\n\n10:00–10:30 am\nAndrea Puhm\nTitle: Celestial wave scattering on Kerr-Schild backgrounds\n\n\n10:30–11:00 am\nCoffee break\n\n\n\n11:00–11:30 am\nSabrina Pasterski\nTitle: Mining Celestial Symmetries \nAbstract: The aim of this talk is to delve into the common thread that ties together recent work with H. Verlinde\, L. Donnay\, A. Puhm\, and S. Banerjee exploring\, explaining\, and exploiting the symmetries encoded in the conformally soft sector. \nCome prepared to debate the central charge\, loop corrections\, contour prescriptions\, and orders of limits!\n\n\n11:30–12:00 pm\nShamik Banerjee\nTitle: Virasoro and other symmetries in CCFT \nAbstract:  In this talk I will briefly describe my ongoing work with Sabrina Pasterski. In this work we revisit the standard construction of the celestial stress tensor as a shadow of the subleading conformally soft graviton.  In its original formulation\, we find that there is an obstruction to reproducing the expected $TT$ OPE in the double soft limit. This obstruction is related to the existence of the $SL_2$ current algebra symmetry of the CCFT. We propose a modification to the definition of the stress tensor which circumvents this obstruction and also discuss its implications for the existence of other current algebra (w_{1+\infty}) symmetries in CCFT.\n\n\n12:00–2:00 pm\nLunch break\n\n\n\n\nAfternoon Session\nChair: Albert Law\n\n\n2:00–2:30 pm\nTomasz Taylor\nTitle: Celestial Yang-Mills amplitudes and D=4 conformal blocks\n\n\n2:30–3:00 pm\nBin Zhu\nTitle:  Single-valued correlators and Banerjee-Ghosh equations \nAbstract:  Low-point celestial amplitudes are plagued with singularities resulting from spacetime translation. We consider a marginal deformation of the celestial CFT which is realized by coupling Yang-Mills theory to a background dilaton field\, with the (complex) dilaton source localized on the celestial sphere. This picture emerges from the physical interpretation of the solutions of the system of differential equations discovered by Banerjee and Ghosh. We show that the solutions can be written as Mellin transforms of the amplitudes evaluated in such a dilaton background. The resultant three-gluon and four-gluon amplitudes are single-valued functions of celestial coordinates enjoying crossing symmetry and all other properties expected from standard CFT correlators.\n\n\n3:00–3:30 pm\nCoffee break\n\n\n\n3:30–4:00 pm\nAlex Lupsasca\nTitle: Holography of the Photon Ring\n\n\n4:00–5:30 pm\nShort Talks\nElizabeth Himwich: Celestial OPEs and w(1+infinity) symmetry of massless and massive amplitudes \nAdam Ball: Perturbatively exact $w_{1+\infty}$ asymptotic symmetry of quantum self-dual gravity \nRomain Ruzziconi: A Carrollian Perspective on Celestial Holography \nJordan Cotler: Soft Gravitons in 3D \nAlfredo Guevara: Comments on w_1+\inf \nAndrew Strominger: Top-down celestial holograms \nEduardo Casali: Celestial amplitudes as AdS-Witten diagrams \nWalker Melton: Top-down celestial holograms\n\n\n\n\n  \nFriday\, June 24\, 2022 \n\n\n\n\n9:00–9:30 am\nBreakfast\n\n\n9:30–12:30 pm\nOpen Discussion\n\n\n12:30–2:30 pm\nLunch provided at the BHI\n\n\n\nDeparture\n\n\n\n\n 
URL:https://cmsa.fas.harvard.edu/event/joint-bhi-cmsa-conference-on-flat-holography/
LOCATION:CMSA Room G10\, CMSA\, 20 Garden Street\, Cambridge\, MA\, 02138\, United States
CATEGORIES:Conference,Event
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/Flat-Holography_2022_small.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220623T090000
DTEND;TZID=America/New_York:20220623T100000
DTSTAMP:20260409T125528
CREATED:20240214T091046Z
LAST-MODIFIED:20240301T101920Z
UID:10002611-1655974800-1655978400@cmsa.fas.harvard.edu
SUMMARY:Some new algorithms in statistical genomics
DESCRIPTION:Abstract: The statistical analysis of genomic data has incubated many innovations for computational method development. This talk will discuss some simple algorithms that may be useful in analyzing such data. Examples include algorithms for efficient resampling-based hypothesis testing\, minimizing the sum of truncated convex functions\, and fitting equality-constrained lasso problems. These algorithms have the potential to be used in other applications beyond statistical genomics. \nBio: Hui Jiang is an Associate Professor in the Department of Biostatistics at the University of Michigan. He received his Ph.D. in Computational and Mathematical Engineering from Stanford University. Before joining the University of Michigan\, he was a postdoc in the Department of Statistics and Stanford Genome Technology Center at Stanford University. He is interested in developing statistical and computational methods for analyzing large-scale biological data generated using modern high-throughput technologies.
URL:https://cmsa.fas.harvard.edu/event/6-23-2022-interdisciplinary-science-seminar/
LOCATION:MA
CATEGORIES:Interdisciplinary Science Seminar
ATTACH;FMTTYPE=image/jpeg:https://cmsa.fas.harvard.edu/media/CMSA-Interdisciplinary-Science-Seminar-06.23.2022-1583x2048-1.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220630T162300
DTEND;TZID=America/New_York:20220630T172300
DTSTAMP:20260409T125528
CREATED:20240214T091304Z
LAST-MODIFIED:20240301T101730Z
UID:10002613-1656606180-1656609780@cmsa.fas.harvard.edu
SUMMARY:Entanglement and its key role in quantum information
DESCRIPTION:Abstract: Entanglement is a type of correlation found in composite quantum systems\, connected with various non-classical phenomena. Currently\, entanglement plays a key role in quantum information applications such as quantum computing\, quantum communication\, and quantum sensing. In this talk the concept of entanglement will be introduced along with various methods that have been proposed to detect and quantify it. The fundamental role of entanglement in both quantum theory and quantum technology will also be discussed. \nBio: Spyros Tserkis is a postdoctoral researcher at Harvard University\, working on quantum information theory. Before joining Harvard in Fall 2021\, he was a postdoctoral researcher at MIT and the Australian National University. He received his PhD from the University of Queensland.
URL:https://cmsa.fas.harvard.edu/event/6-30-2022-interdisciplinary-science-seminar/
LOCATION:MA
CATEGORIES:Interdisciplinary Science Seminar
ATTACH;FMTTYPE=image/jpeg:https://cmsa.fas.harvard.edu/media/CMSA-Interdisciplinary-Science-Seminar-06.30.22-1583x2048-1.jpg
END:VEVENT
END:VCALENDAR