New memory-query tradeoffs for convex optimization algorithms.
problem Optimizing memory usage for convex optimization algorithms.
method Analyzing randomized first-order algorithms for minimizing convex functions.
result Cutting plane methods are optimal in terms of memory and query complexity.
Memory-constrained algorithms need superlinear memory for efficient convex optimization.
problem Efficiently minimizing convex functions with limited memory.
method Analyzing first-order algorithms with superlinear memory constraints.
result Superlinear memory is necessary for optimal performance in convex optimization.
A technique identifies memoryless algorithms approximating memory-dependent optimization methods.
problem Understanding how memory in optimization algorithms affects loss and generalization.
method Introducing a general technique to replace past iterates with the current one and adding a correction term.
result Lion does not have the same implicit anti-regularization as AdamW, explaining its better generalization performance.
New analysis tightens memory capacity of Hopfield models using spherical codes.
problem Optimizing memory capacity in modern Hopfield models and Kernelized Hopfield Models.
method Connecting Hopfield models to spherical codes in information theory, establishing an optimal capacity bound and a sub-linear algorithm.
result First tight and optimal asymptotic memory capacity for modern Hopfield models, matching known lower bounds.
Paper tackles online optimization with memory and competitive control.
problem Minimizing hitting and switching costs in online optimization problems.
method Optimistic Regularized Online Balanced Descent algorithm.
result Achieves a constant, dimension-free competitive ratio.
This work analyzes and optimizes memory and compute costs of learned optimizers.
problem High memory and compute costs of learned optimizers.
method Identified and quantified design features of learned and hand-designed optimizers, constructed a more efficient learned optimizer.
result A learned optimizer that is faster and more memory efficient than previous work.
Quadratic memory is essential for optimal convex optimization queries.
problem Optimal query complexity for convex optimization and feasibility problems.
method Lower bounds on query complexity for convex optimization and feasibility problems.
result Center-of-mass algorithms are Pareto-optimal for both convex optimization and feasibility problems.
State-of-the-art models are now trained with billions of parameters, reaching hardware limits in terms of memory consumption. This has created a recent demand for memory-efficient optimizers. To this end, we investigate the limits and performance tradeoffs of memory-efficient adaptively preconditioned gradient methods.…
Paper proposes efficient optimizers for large language models with fast convergence and low memory usage.
problem Designing efficient optimizers for large language models with low-memory requirements and fast convergence.
method Structured Fisher information matrix approximation and low-rank extension framework.
result New optimizers (RACS and Alice) achieve better convergence and lower memory usage than existing methods.
We note that known methods achieving the optimal oracle complexity for first order convex optimization require quadratic memory, and ask whether this is necessary, and more broadly seek to characterize the minimax number of first order queries required to optimize a convex Lipschitz function subject to a memory constra…
Memory-Augmented Meta-Optimization improves cold-start recommendation.
problem Cold-start problem in recommender systems for new users or items.
method Memory-Augmented Meta-Optimization approach with personalized and task-specific memories.
result Significant improvement in cold-start recommendation performance on multiple datasets.
Training machine learning models on mobile devices has the potential of improving both privacy and accuracy of the models. However, one of the major obstacles to achieving this goal is the memory limitation of mobile devices. Reducing training memory enables models with high-dimensional weight matrices, like automatic …
TASO optimizes CNN models for memory-constrained devices.
problem Large CNN models are too resource-intensive for mobile devices.
method Integer Linear Programming (ILP) for selecting operations and workspace allocation.
result 8x speedup and 2.2x reduction in memory usage compared to greedy algorithms.
New algorithms for constrained online optimization with memory and predictions.
problem Control of constrained dynamical systems and scheduling with reconfiguration budgets.
method Proposed algorithms achieving sublinear regret and constraint violation under time-varying constraints, both with and without predictions.
result First algorithms achieving sublinear regret and constraint violation in constrained online optimization with memory.
Study on memory effects in RNNs learning temporal data.
problem Understanding memory effects in RNNs for temporal data learning.
method Mathematical analysis of continuous-time linear RNNs, focusing on approximation and optimization dynamics.
result Long-term memory requires a large number of neurons and slows down training.
Second-order optimizers retain residual information after data deletion, affecting machine unlearning.
problem Residual information in second-order optimizers after data deletion.
method Comparison of first-order and second-order learners, eigendecomposition analysis.
result Second-order optimizers retain residual information, not detectable by first-order analysis.
Serenity optimizes neural network execution for edge devices by scheduling with optimal memory footprint.
problem Order of nodes in irregular neural networks affects memory footprint, complicating execution under resource constraints.
method Memory-aware compiler using dynamic programming and graph rewriting to find optimal schedules.
result Achieves optimal peak memory and further improves it with graph rewriting, reducing memory usage by 1.68x-1.86x compared to TensorFlow Lite.
LDAdam optimizes large models with low memory by adapting to lower-dimensional subspaces.
problem Training large models efficiently and accurately.
method Adaptive optimization in lower-dimensional subspaces with a new projection-aware update rule and error feedback mechanism.
result LDAdam achieves accurate and efficient training of language models.
Optimal CL requires perfect memory and is NP-hard.
problem Designing CL algorithms that perform reliably and avoid catastrophic forgetting.
method Theoretical approach to derive computational properties of optimal CL algorithms.
result Optimal CL algorithms generally solve an NP-hard problem and require perfect memory.
Optimal linear contracts are possible even with memory in Gaussian settings.
problem Can optimal dynamic contracts be linear when agents control memory processes?
method Developed a methodology for non-Markovian and non-semimartingale settings, showed linear contracts are optimal for one-dimensional models.
result Linear contracts are optimal for one-dimensional models with memory, and for radial effort cost functions in higher dimensions.
Memory-efficient algorithm reduces variance in off-policy RL.
problem High variance in off-policy policy optimization.
method Memory-efficient, stochastically variance-reduced algorithm using off-policy samples.
result Empirically validated effectiveness of the proposed algorithm.
Sketchy reduces memory and compute requirements for adaptive regularization in deep learning.
problem Prohibitive memory and running time for adaptive regularization methods in deep learning.
method Low-rank sketching approach using Frequent Directions (FD) to reduce memory and compute requirements.
result Efficient interpolation between resource requirements and degradation in regret guarantees with rank k k k . AlphaGrad optimizes memory usage in RL algorithms by normalizing gradients.
problem Memory overhead and hyperparameter complexity in adaptive optimizers.
method Tensor-wise L2 normalization followed by a smooth hyperbolic tangent transformation controlled by a single parameter.
result AlphaGrad provides enhanced training stability and competitive performance in various RL algorithms.
New algorithms solve convex optimization problems with limited memory.
problem Solving convex optimization problems with constrained memory.
method Recursive cutting-plane algorithms dividing variables into blocks.
result Achieves optimal memory usage and oracle complexity in certain regimes.
A new memory-efficient Adam variant reduces second moments when feasible.
problem Memory constraints in training machine learning models.
method Signal-to-Noise Ratio (SNR) analysis to identify dimensions where second moments can be replaced by means.
result Memory-efficient Adam variant (SlimAdam) matches performance and stability of Adam while saving up to 98% of second moments.
OpTorch optimizes deep learning for resource-limited environments.
problem Resource constraints in deep learning training.
method Optimized deep learning pipelines in training time and memory.
result Achieved similar accuracy to existing libraries with reduced memory usage.
Deep Neural Networks(DNNs) require huge GPU memory when training on modern image/video databases. Unfortunately, the GPU memory is physically finite, which limits the image resolutions and batch sizes that could be used in training for better DNN performance. Unlike solutions that require physically upgrade GPUs, the G…
NEST optimizes deep learning training by placing devices efficiently across networks and memory.
problem Inefficient device placement in distributed deep learning leads to high communication and memory overhead.
method NEST uses network-, compute-, and memory-aware dynamic programming to optimize device placement.
result NEST achieves up to 2.43 times higher throughput and better memory efficiency.
Memory-based models can learn to approximate Bayes-optimal predictors for non-stationary data.
problem Learning from non-stationary data with unobserved switching points.
method Memory-based neural models, including Transformers, LSTMs, and RNNs, trained to minimize log loss.
result Memory-based models can accurately approximate known Bayes-optimal algorithms and perform Bayesian inference over latent switching points.
This study examines memory effects in S&P500 market correlations using Langevin models.
problem The neglect of memory effects in market correlations for optimal portfolio selection.
method Fit a generalised Langevin equation (GLE) to S&P500 market correlation data.
result Memory effects in market correlations significantly improve forecasting accuracy and suggest a hidden slow time scale.
We present an end-to-end trained memory system that quickly adapts to new data and generates samples like them. Inspired by Kanerva's sparse distributed memory, it has a robust distributed reading and writing mechanism. The memory is analytically tractable, which enables optimal on-line compression via a Bayesian updat…
New algorithms achieve optimal regret in sliding window model with limited memory.
problem Experts problem in the sliding window model with limited information.
method 2 queries, polylog(nT) memory, exponential improvement on memory.
result Achieve optimal regret of sqrt(nW)polylog(nT) with 2 queries and polylog(nT) memory.
We propose two optimization techniques to minimize memory usage and computation while meeting system timing constraints for real-time classification in wearable systems. Our method derives a hierarchical classifier structure for Support Vector Machine (SVM) in order to reduce the amount of computations, based on the pr…
POET enables large neural network training on tiny devices with reduced energy.
problem Training large neural networks on memory-limited edge devices.
method Jointly optimizes rematerialization and paging for memory reduction, formulating an MILP for energy-efficient training.
result POET trains ResNet-18 and BERT within Cortex-M memory constraints, outperforming current methods in energy efficiency.
SMGD trains low-bit neural networks with memory constraints.
problem Training large neural networks with limited memory.
method Stochastic Markov Gradient Descent (SMGD).
result Encouraging numerical results and theoretical guarantees.
We present Memory Augmented Policy Optimization (MAPO), a simple and novel way to leverage a memory buffer of promising trajectories to reduce the variance of policy gradient estimate. MAPO is applicable to deterministic environments with discrete actions, such as structured prediction and combinatorial optimization ta…
Memory-augmented neural networks consisting of a neural controller and an external memory have shown potentials in long-term sequential learning. Current RAM-like memory models maintain memory accessing every timesteps, thus they do not effectively leverage the short-term memory held in the controller. We hypothesize t…
This paper optimizes SMPC for neural network inference, reducing memory and time.
problem Memory and time constraints in secure neural network inference.
method Implemented ABY2.0 protocol, optimized memory usage, and used a helper node.
result MNIST inference reduced from 8.03 GB RAM and 200s to 0.2 GB RAM and 32s.
Efficiently discovers Bayesian network structure with reduced memory usage.
problem Bayesian network structure discovery is NP-hard and memory-intensive.
method Progressively leveled scoring approach with hierarchical computation.
result Achieved processing of a 28-variable Bayesian network using only memory.
New method improves FO-BLO convergence without increasing memory or time complexity.
problem Lack of theoretical understanding of FO-BLO convergence.
method Unbiased first-order bilevel optimization (UFO-BLO) to improve convergence.
result Theoretical guarantee of convergence for FO-BLO-based stochastic optimization.
New framework captures long-term decision dependence in online learning.
problem Long-term dependence on past decisions in online learning.
method Introduces Online Convex Optimization with Unbounded Memory (OCO-UMB) and p p p -effective memory capacity. result Proves O ( H p T ) O(\sqrt{H_p T}) O ( H p T ) upper bound on policy regret and matching lower bound. Gradient descent is efficient for solving feasibility problems with minimal memory and queries.
problem Finding a point in a given set using a memory-constrained algorithm with a separation oracle.
method Oracle complexity lower bounds for gradient descent and other algorithms.
result Gradient descent is Pareto-optimal in the oracle complexity/memory tradeoff for feasibility problems.
In this paper, the Kyle model of insider trading is extended by characterizing the trading volume with long memory and allowing the noise trading volatility to follow a general stochastic process. Under this newly revised model, the equilibrium conditions are determined, with which the optimal insider trading strategy,…
HiPPO framework optimizes memory compression for sequential data.
problem Incremental representation of cumulative history in sequential data.
method Optimal polynomial projections for online function approximation.
result HiPPO-LegS achieves state-of-the-art accuracy on MNIST.
A new method reduces memory requirements for sorting high-dimensional data.
problem Efficiently sorting and organizing high-dimensional data with low memory usage.
method Iteratively shuffles N indices and applies SoftSort optimization steps.
result Significantly improves sorting quality for multidimensional data.
Adaptive gradient-based optimizers such as Adagrad and Adam are crucial for achieving state-of-the-art performance in machine translation and language modeling. However, these methods maintain second-order statistics for each parameter, thus introducing significant memory overheads that restrict the size of the model b…
The paper studies scaling laws for associative memory mechanisms.
problem Understanding and optimizing learning and memorization processes.
method High-dimensional matrices of outer products of embeddings, relating to transformer models. Derived scaling laws with sample and parameter sizes. Extensive numerical experiments.
result Precise scaling laws and statistical efficiency of estimators.
The key to a Transformer model is the self-attention mechanism, which allows the model to analyze an entire sequence in a computationally efficient manner. Recent work has suggested the possibility that general attention mechanisms used by RNNs could be replaced by active-memory mechanisms. In this work, we evaluate wh…