Ali Shameli

Youwei Liang, Kevin Stone, Ali Shameli et al.

Finding the optimal pass sequence of compilation can lead to a significant reduction in program size and/or improvement in program efficiency. Prior works on compilation pass ordering have two major drawbacks. They either require an excessive budget (in terms of compilation steps) at compile time or fail to generalize to unseen programs. In this paper, for code-size reduction tasks, we propose a novel pipeline to find program-dependent pass sequences within 45 compilation calls. It first identifies a coreset of 50 pass sequences via greedy optimization of a submodular function, and then learns a policy with Graph Neural Network (GNN) to pick the optimal sequence by predicting the normalized values of the pass sequences in the coreset. Despite its simplicity, our pipeline outperforms the default -Oz flag by an average of 4.7% over a large collection (4683) of unseen code repositories from diverse domains across 14 datasets. In comparison, previous approaches like reinforcement learning on the raw pass sequence space may take days to train due to sparse reward, and may not generalize well in held-out ones from different domains. Our results demonstrate that existing human-designed compiler flags can be improved with a simple yet effective technique that transforms the raw action space into a small one with denser rewards.

Prem Talwai, Ali Shameli, David Simchi-Levi

We develop novel learning rates for conditional mean embeddings by applying the theory of interpolation for reproducing kernel Hilbert spaces (RKHS). We derive explicit, adaptive convergence rates for the sample estimator under the misspecifed setting, where the target operator is not Hilbert-Schmidt or bounded with respect to the input/output RKHSs. We demonstrate that in certain parameter regimes, we can achieve uniform convergence rates in the output RKHS. We hope our analyses will allow the much broader application of conditional mean embeddings to more complex ML/RL settings involving infinite dimensional RKHSs and continuous state spaces.

Tan Nguyen, Ali Shameli, Yasin Abbasi-Yadkori et al.

We study sample complexity of optimizing "hill-climbing friendly" functions defined on a graph under noisy observations. We define a notion of convexity, and we show that a variant of best-arm identification can find a near-optimal solution after a small number of queries that is independent of the size of the graph. For functions that have local minima and are nearly convex, we show a sample complexity for the classical simulated annealing under noisy observations. We show effectiveness of the greedy algorithm with restarts and the simulated annealing on problems of graph-based nearest neighbor classification as well as a web document re-ranking application.

Ali Shameli, Yasin Abbasi-Yadkori

The continuation method is a popular approach in non-convex optimization and computer vision. The main idea is to start from a simple function that can be minimized efficiently, and gradually transform it to the more complicated original objective function. The solution of the simpler problem is used as the starting point to solve the original problem. We show a continuation method for discrete optimization problems. Ideally, we would like the evolved function to be hill-climbing friendly and to have the same global minima as the original function. We show that the proposed continuation method is the best affine approximation of a transformation that is guaranteed to transform the function to a hill-climbing friendly function and to have the same global minima. We show the effectiveness of the proposed technique in the problem of nearest neighbor classification. Although nearest neighbor methods are often competitive in terms of sample efficiency, the computational complexity in the test phase has been a major obstacle in their applicability in big data problems. Using the proposed continuation method, we show an improved graph-based nearest neighbor algorithm. The method is readily understood and easy to implement. We show how the computational complexity of the method in the test phase scales gracefully with the size of the training set, a property that is particularly important in big data applications.

Ali Shameli, Tim Althoff, Amin Saberi et al.

Gamification represents an effective way to incentivize user behavior across a number of computing applications. However, despite the fact that physical activity is essential for a healthy lifestyle, surprisingly little is known about how gamification and in particular competitions shape human physical activity. Here we study how competitions affect physical activity. We focus on walking challenges in a mobile activity tracking application where multiple users compete over who takes the most steps over a predefined number of days. We synthesize our findings in a series of game and app design implications. In particular, we analyze nearly 2,500 physical activity competitions over a period of one year capturing more than 800,000 person days of activity tracking. We observe that during walking competitions, the average user increases physical activity by 23%. Furthermore, there are large increases in activity for both men and women across all ages, and weight status, and even for users that were previously fairly inactive. We also find that the composition of participants greatly affects the dynamics of the game. In particular, if highly unequal participants get matched to each other, then competition suffers and the overall effect on the physical activity drops significantly. Furthermore, competitions with an equal mix of both men and women are more effective in increasing the level of activities. We leverage these insights to develop a statistical model to predict whether or not a competition will be particularly engaging with significant accuracy. Our models can serve as a guideline to help design more engaging competitions that lead to most beneficial behavioral changes.