More Like this

1. Explicit Mean-Square Error Bounds for Monte-Carlo and Linear Stochastic Approximation

   Chen, Shuhang and (July 2020, Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics)

   This paper concerns error bounds for recursive equations subject to Markovian disturbances. Motivating examples abound within the fields of Markov chain Monte Carlo (MCMC) and Reinforcement Learning (RL), and many of these algorithms can be interpreted as special cases of stochastic approximation (SA). It is argued that it is not possible in general to obtain a Hoeffding bound on the error sequence, even when the underlying Markov chain is reversible and geometrically ergodic, such as the M/M/1 queue. This is motivation for the focus on mean square error bounds for parameter estimates. It is shown that mean square error achieves the optimal rate of $O(1/n)$, subject to conditions on the step-size sequence. Moreover, the exact constants in the rate are obtained, which is of great value in algorithm design 

   more » « less
2. Momentum-Based Variance Reduction in Non-Convex SGD

   Cutkosky, Ashok; Orabona, Francesco (January 2019, Advances in neural information processing systems)

   Wallach, H.; Larochelle, H.; Beygelzimer, A.; d'Alché-Buc, F.; Fox, E.; Garnett, R. (Ed.)

   Variance reduction has emerged in recent years as a strong competitor to stochastic gradient descent in non-convex problems, providing the first algorithms to improve upon the converge rate of stochastic gradient descent for finding first-order critical points. However, variance reduction techniques typically require carefully tuned learning rates and willingness to use excessively large "mega-batches" in order to achieve their improved results. We present a new algorithm, STORM, that does not require any batches and makes use of adaptive learning rates, enabling simpler implementation and less hyperparameter tuning. Our technique for removing the batches uses a variant of momentum to achieve variance reduction in non-convex optimization. On smooth losses $$F$$, STORM finds a point $$\boldsymbol{x}$$ with $$E[\|\nabla F(\boldsymbol{x})\|]\le O(1/\sqrt{T}+\sigma^{1/3}/T^{1/3})$$ in $$T$$ iterations with $$\sigma^2$$ variance in the gradients, matching the optimal rate and without requiring knowledge of $$\sigma$$. 

   more » « less
3. Statistical and Computational Complexities of BFGS Quasi-Newton Method for Generalized Linear Models

   Jin, Qiujiang; Ren, Tongzheng; Ho, Nhat; Mokhtari, Aryan (May 2024, Transactions on machine learning research)

   The gradient descent (GD) method has been used widely to solve parameter estimation in generalized linear models (GLMs), a generalization of linear models when the link function can be non-linear. In GLMs with a polynomial link function, it has been shown that in the high signal-to-noise ratio (SNR) regime, due to the problem's strong convexity and smoothness, GD converges linearly and reaches the final desired accuracy in a logarithmic number of iterations. In contrast, in the low SNR setting, where the problem becomes locally convex, GD converges at a slower rate and requires a polynomial number of iterations to reach the desired accuracy. Even though Newton's method can be used to resolve the flat curvature of the loss functions in the low SNR case, its computational cost is prohibitive in high-dimensional settings as it is $$\mathcal{O}(d^3)$$, where $$d$$ the is the problem dimension. To address the shortcomings of GD and Newton's method, we propose the use of the BFGS quasi-Newton method to solve parameter estimation of the GLMs, which has a per iteration cost of $$\mathcal{O}(d^2)$$. When the SNR is low, for GLMs with a polynomial link function of degree $$p$$, we demonstrate that the iterates of BFGS converge linearly to the optimal solution of the population least-square loss function, and the contraction coefficient of the BFGS algorithm is comparable to that of Newton's method. Moreover, the contraction factor of the linear rate is independent of problem parameters and only depends on the degree of the link function $$p$$. Also, for the empirical loss with $$n$$ samples, we prove that in the low SNR setting of GLMs with a polynomial link function of degree $$p$$, the iterates of BFGS reach a final statistical radius of $$\mathcal{O}((d/n)^{\frac{1}{2p+2}})$$ after at most $$\log(n/d)$$ iterations. This complexity is significantly less than the number required for GD, which scales polynomially with $(n/d)$. 

   more » « less
4. MDPGT: Momentum-Based Decentralized Policy Gradient Tracking

   https://doi.org/10.1609/aaai.v36i9.21169  

   Jiang, Zhanhong; Lee, Xian Yeow; Tan, Sin Yong; Tan, Kai Liang; Balu, Aditya; Lee, Young M; Hegde, Chinmay; Sarkar, Soumik (June 2022, Proceedings of the AAAI Conference on Artificial Intelligence)

   We propose a novel policy gradient method for multi-agent reinforcement learning, which leverages two different variance-reduction techniques and does not require large batches over iterations. Specifically, we propose a momentum-based decentralized policy gradient tracking (MDPGT) where a new momentum-based variance reduction technique is used to approximate the local policy gradient surrogate with importance sampling, and an intermediate parameter is adopted to track two consecutive policy gradient surrogates. MDPGT provably achieves the best available sample complexity of O(N -1 e -3) for converging to an e-stationary point of the global average of N local performance functions (possibly nonconcave). This outperforms the state-of-the-art sample complexity in decentralized model-free reinforcement learning and when initialized with a single trajectory, the sample complexity matches those obtained by the existing decentralized policy gradient methods. We further validate the theoretical claim for the Gaussian policy function. When the required error tolerance e is small enough, MDPGT leads to a linear speed up, which has been previously established in decentralized stochastic optimization, but not for reinforcement learning. Lastly, we provide empirical results on a multi-agent reinforcement learning benchmark environment to support our theoretical findings. 

   more » « less
5. Submodular maximization with matroid and packing constraints in parallel

   https://doi.org/10.1145/3313276.3316389  

   Ene, Alina; Nguyen, Huy L.; Vladu, Adrian (June 2019, ACM SIGACT Symposium on Theory of Computing)

   We consider the problem of maximizing the multilinear extension of a submodular function subject a single matroid constraint or multiple packing constraints with a small number of adaptive rounds of evaluation queries. We obtain the first algorithms with low adaptivity for submodular maximization with a matroid constraint. Our algorithms achieve a $$1-1/e-\epsilon$$ approximation for monotone functions and a $$1/e-\epsilon$$ approximation for non-monotone functions, which nearly matches the best guarantees known in the fully adaptive setting. The number of rounds of adaptivity is $$O(\log^2{n}/\epsilon^3)$$, which is an exponential speedup over the existing algorithms. We obtain the first parallel algorithm for non-monotone submodular maximization subject to packing constraints. Our algorithm achieves a $$1/e-\epsilon$$ approximation using $$O(\log(n/\epsilon) \log(1/\epsilon) \log(n+m)/ \epsilon^2)$$ parallel rounds, which is again an exponential speedup in parallel time over the existing algorithms. For monotone functions, we obtain a $$1-1/e-\epsilon$$ approximation in $$O(\log(n/\epsilon)\log(m)/\epsilon^2)$$ parallel rounds. The number of parallel rounds of our algorithm matches that of the state of the art algorithm for solving packing LPs with a linear objective (Mahoney et al., 2016). Our results apply more generally to the problem of maximizing a diminishing returns submodular (DR-submodular) function. 

   more » « less