Classical probabilistic graphical models face fundamental challenges in modern data environments, which are characterized by high dimensionality, source heterogeneity, and stringent data-sharing constraints. In this work, we revisit the Ising model, a well-established member of the Markov Random Field (MRF) family, and develop a distributed framework that enables scalable and privacy-preserving representation learning from large-scale binary data with inherent low-rank structure. Our approach optimizes a non-convex surrogate loss function via bi-factored gradient descent, offering substantial computational and communication advantages over conventional convex approaches. We evaluate our algorithm on multi-institutional electronic health record (EHR) datasets from 58,248 patients across the University of Pittsburgh Medical Center (UPMC) and Mass General Brigham (MGB), demonstrating superior performance in global representation learning and downstream clinical tasks, including relationship detection, patient phenotyping, and patient clustering. These results highlight a broader potential for statistical inference in federated, high-dimensional settings while addressing the practical challenges of data complexity and multi-institutional integration.

Keeping risk under control is often more crucial than maximizing expected rewards in real-world decision-making situations, such as finance, robotics, autonomous driving, etc. The most natural choice of risk measures is variance, which penalizes the upside volatility as much as the downside part. Instead, the (downside) semivariance, which captures the negative deviation of a random variable under its mean, is more suitable for risk-averse proposes. This paper aims at optimizing the mean-semivariance (MSV) criterion in reinforcement learning w.r.t. steady reward distribution. Since semivariance is time-inconsistent and does not satisfy the standard Bellman equation, the traditional dynamic programming methods are inapplicable to MSV problems directly. To tackle this challenge, we resort to Perturbation Analysis (PA) theory and establish the performance difference formula for MSV. We reveal that the MSV problem can be solved by iteratively solving a sequence of RL problems with a policy-dependent reward function. Further, we propose two on-policy algorithms based on the policy gradient theory and the trust region method. Finally, we conduct diverse experiments from simple bandit problems to continuous control tasks in MuJoCo, which demonstrate the effectiveness of our proposed methods.

Dynamic optimization of mean and variance in Markov decision processes (MDPs) is a long-standing challenge caused by the failure of dynamic programming. In this paper, we propose a new approach to find the globally optimal policy for combined metrics of steady-state mean and variance in an infinite-horizon undiscounted MDP. By introducing the concepts of pseudo mean and pseudo variance, we convert the original problem to a bilevel MDP problem, where the inner one is a standard MDP optimizing pseudo mean-variance and the outer one is a single parameter selection problem optimizing pseudo mean. We use the sensitivity analysis of MDPs to derive the properties of this bilevel problem. By solving inner standard MDPs for pseudo mean-variance optimization, we can identify worse policy spaces dominated by optimal policies of the pseudo problems. We propose an optimization algorithm which can find the globally optimal policy by repeatedly removing worse policy spaces. The convergence and complexity of the algorithm are studied. Another policy dominance property is also proposed to further improve the algorithm efficiency. Numerical experiments demonstrate the performance and efficiency of our algorithms. To the best of our knowledge, our algorithm is the first that efficiently finds the globally optimal policy of mean-variance optimization in MDPs. These results are also valid for solely minimizing the variance metrics in MDPs.

Keeping risk under control is often more crucial than maximizing expected reward in real-world decision-making situations, such as finance, robotics, autonomous driving, etc. The most natural choice of risk measures is variance, while it penalizes the upside volatility as much as the downside part. Instead, the (downside) semivariance, which captures the negative deviation of a random variable under its mean, is more suitable for risk-averse proposes. This paper aims at optimizing the mean-semivariance (MSV) criterion in reinforcement learning w.r.t. steady rewards. Since semivariance is time-inconsistent and does not satisfy the standard Bellman equation, the traditional dynamic programming methods are inapplicable to MSV problems directly. To tackle this challenge, we resort to the Perturbation Analysis (PA) theory and establish the performance difference formula for MSV. We reveal that the MSV problem can be solved by iteratively solving a sequence of RL problems with a policy-dependent reward function. Further, we propose two on-policy algorithms based on the policy gradient theory and the trust region method. Finally, we conduct diverse experiments from simple bandit problems to continuous control tasks in MuJoCo, which demonstrate the effectiveness of our proposed methods.

Late last week, Google research scientist Fei Xia sat in the center of a bright, open-plan kitchen and typed a command into a laptop connected to a one-armed, wheeled robot resembling a large floor lamp. The robot promptly zoomed over to a nearby countertop, gingerly picked up a bag of multigrain chips with a large plastic pincer, and wheeled over to Xia to offer up a snack. The most impressive thing about that demonstration, held in Google's robotics lab in Mountain View, California, was that no human coder had programmed the robot to understand what to do in response to Xia's command. Its control software had learned how to translate a spoken phrase into a sequence of physical actions using millions of pages of text scraped from the web. That means a person doesn't have to use specific preapproved wording to issue commands, as can be necessary with virtual assistants such as Alexa or Siri.

Late last week, Google research scientist Fei Xia sat in the center of a bright, open-plan kitchen and typed a command into a laptop connected to a one-armed, wheeled robot resembling a large floor lamp. The robot promptly zoomed over to a nearby countertop, gingerly picked up a bag of multigrain chips with a large plastic pincer, and wheeled over to Xia to offer up a snack. The most impressive thing about that demonstration, held in Google's robotics lab in Mountain View, California, was that no human coder had programmed the robot to understand what to do in response to Xia's command. Its control software had learned how to translate a spoken phrase into a sequence of physical actions using millions of pages of text scraped from the web. That means a person doesn't have to use specific preapproved wording to issue commands, as can be necessary with virtual assistants such as Alexa or Siri.

When agents have conflicting preferences over a set of alternatives and they want to make a joint decision, a natural way to do so is by voting. How to design and analyze desirable voting rules has been studied by economists for centuries. In recent decades, technological advances, especially those in internet economy, have introduced many new applications for voting theory. For example, we can rate movies based on people's preferences, as done on many movie recommendation sites. However, in such new applications, we always encounter a large number of alternatives or an overwhelming amount of information, which makes computation in voting process a big challenge. Such challenges have led to a burgeoning area--computational social choice, aiming to address problems in computational aspects of preference representation and aggregation in a multi-agent scenario. The high-level goal of my research is to better understand and prevent the agents' (strategic) behavior in voting systems, as well as to design computationally efficient ways for agents to present their preferences and make a joint decision.

Last year, Charlene Xia '17, SM '20 found herself at a crossroads. She was finishing up her master's degree in media arts and sciences from the MIT Media Lab and had just submitted applications to doctoral degree programs. All Xia could do was sit and wait. In the meantime, she narrowed down her career options, regardless of whether she was accepted to any program. "I had two thoughts: I'm either going to get a PhD to work on a project that protects our planet, or I'm going to start a restaurant," recalls Xia.

Last year, Charlene Xia '17, SM '20 found herself at a crossroads. She was finishing up her master's degree in media arts and sciences from the MIT Media Lab and had just submitted applications to doctoral degree programs. All Xia could do was sit and wait. In the meantime, she narrowed down her career options, regardless of whether she was accepted to any program. "I had two thoughts: I'm either going to get a PhD to work on a project that protects our planet, or I'm going to start a restaurant," recalls Xia.

Zao, a free deepfake face-swapping app that's able to place your likeness into scenes from hundreds of movies and TV shows after uploading just a single photograph, has gone viral in China. Bloomberg reports that the app was released on Friday, and quickly reached the top of the free charts on the Chinese iOS App Store. And like the FaceApp aging app before it, the creators of Zao are now facing a backlash over a perceived threat to user privacy. Twitter user Allan Xia posted a neat demonstration of what the app is capable of yesterday with a 30 second clip of their face replacing Leonardo Dicaprio in famous moments from several of his films. According to Xia, the clips were generated in under eight seconds from just a single photograph, however Bloomberg notes that the app can also guide you through the process of taking a series of photographs -- where it will ask you to open and close your mouth and eyes -- to generate more realistic results.