MBZUAI researchers presented a new strategy for handling complex optimization problems in machine learning at ICLR 2024. The study, a collaboration with ISAM, combines zeroth-order methods with hard-thresholding to address specific settings in machine learning. This approach aims to improve convergence, ensuring algorithms reach quality solutions efficiently. Why it matters: Improving optimization techniques is crucial for advancing machine learning models used in various applications, potentially accelerating development and enhancing performance.
Francesco Orabona from Boston University, with a PhD from the University of Genova, researches online learning, optimization, and statistical learning theory. He previously worked at Yahoo Labs and Toyota Technological Institute at Chicago. MBZUAI hosted a panel discussion (topic not specified in provided text). Why it matters: Optimization algorithms are crucial for advancing machine learning and AI, and researchers like Orabona contribute to this field.
MBZUAI and KAUST researchers collaborated to present new optimization methods at ICML 2024 for composite and distributed machine learning settings. The study addresses challenges in training large models due to data size and computational power. Their work focuses on minimizing the "loss function" by adjusting internal trainable parameters, using techniques like gradient clipping. Why it matters: This research contributes to the ongoing advancement of machine learning optimization, crucial for improving the performance and efficiency of AI models in the region and globally.
MBZUAI PhD graduate William de Vazelhes is researching hard-thresholding algorithms to enable AI to work from smaller datasets. His work focuses on optimization algorithms that simplify data, making it easier to analyze and work with, useful for energy-saving and deploying AI models on low-memory devices. He demonstrated that his approach can obtain results similar to those of convex algorithms in many usual settings. Why it matters: This research could broaden AI accessibility by reducing computational costs, and has potential applications in sectors like finance, particularly for portfolio management under budgetary constraints.
Alexander Gasnikov from the Moscow Institute of Physics and Technology presented a talk on open problems in convex optimization. The talk covered stochastic averaging vs stochastic average approximation, saddle-point problems and accelerated methods, homogeneous federated learning, and decentralized optimization. Gasnikov's research focuses on optimization algorithms and he has published in NeurIPS, ICML, EJOR, OMS, and JOTA. Why it matters: While the talk itself isn't directly related to GCC AI, understanding convex optimization is crucial for advancing machine learning algorithms used in the region.
This article discusses approximating a high-dimensional distribution using Gaussian variational inference by minimizing Kullback-Leibler divergence. It builds upon previous research and approximates the minimizer using a Gaussian distribution with specific mean and variance. The study details approximation accuracy and applicability using efficient dimension, relevant for analyzing sampling schemes in optimization. Why it matters: This theoretical research can inform the development of more efficient and accurate AI algorithms, particularly in areas dealing with high-dimensional data such as machine learning and data analysis.
MBZUAI researchers presented a new second-order method for optimizing neural networks at NeurIPS 2024. The method addresses optimization problems related to variational inequalities common in machine learning. They demonstrated that for monotone inequalities with inexact second-order derivatives, no faster second- or first-order methods can theoretically exist, supporting this with experiments. Why it matters: This research has the potential to reduce the computational cost of training large and complex neural networks, which could accelerate AI development in the region.
Mladen Kolar from the University of Chicago Booth School of Business discussed stochastic optimization with equality constraints at MBZUAI. He presented a stochastic algorithm based on sequential quadratic programming (SQP) using a differentiable exact augmented Lagrangian. The algorithm adapts random stepsizes using a stochastic line search procedure, establishing global "almost sure" convergence. Why it matters: The presentation highlights MBZUAI's role in hosting discussions on advanced optimization techniques, fostering research and knowledge exchange in the field of machine learning.