Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

169,051 papers · 148 categories

Trend · papers per month

103206308411 · Jun 202019922001200920182026
48 results for Reduced Precision

Paper proposes training deep neural networks with 8-bit floating point precision.

problem Challenges in training deep neural networks at 8-bit precision due to higher precision and dynamic range requirements.
method Proposes a method to train deep neural networks using 8-bit floating point for weights, activations, errors, and gradients. Introduces an enhanced loss scaling method and stochastic rounding technique.
result Demonstrates state-of-the-art accuracy across multiple datasets and workloads compared to full precision baseline.

New method analyzes accumulation precision in deep learning networks.

problem Lack of precision analysis for accumulation in deep learning training.
method Statistical approach to analyze partial sum accumulations and derive equations for minimum required bits.
result Reduced accumulation precision can lead to loss of information and degraded network quality.

Paper explores low-precision SGLD for neural networks, reducing costs without sacrificing performance.

problem Infeasibility of low-precision sampling in large-scale scenarios.
method Developed low-precision SGLD with quantization function and full-precision gradient accumulators.
result Low-precision SGLD achieves comparable performance to full-precision SGLD with only 8 bits.

Training deep neural networks with 8-bit floating point numbers is now possible and more efficient.

problem Challenges in training DNNs with reduced precision, especially for gradient computations.
method Introduction of chunk-based accumulation and floating point stochastic rounding to reduce arithmetic precision to 16 bits.
result Successful training of DNNs using 8-bit floating point numbers, maintaining accuracy on various models and datasets.

Low-precision computation is often used to lower the time and energy cost of machine learning, and recently hardware accelerators have been developed to support it. Still, it has been used primarily for inference - not training. Previous low-precision training algorithms suffered from a fundamental tradeoff: as the num…

2018-03-09abs ↗pdf ↗

Neural networks with low-precision weights and activations offer compelling efficiency advantages over their full-precision equivalents. The two most frequently discussed benefits of quantization are reduced memory consumption, and a faster forward pass when implemented with efficient bitwise operations. We propose a t…

2017-11-01abs ↗pdf ↗

Reduced numerical precision is a common technique to reduce computational cost in many Deep Neural Networks (DNNs). While it has been observed that DNNs are resilient to small errors and noise, no general result exists that is capable of predicting a given DNN system architecture's sensitivity to reduced precision. In …

2018-05-03abs ↗pdf ↗

Similar to convolution neural networks, recurrent neural networks (RNNs) typically suffer from over-parameterization. Quantizing bit-widths of weights and activations results in runtime efficiency on hardware, yet it often comes at the cost of reduced accuracy. This paper proposes a quantization approach that increases…

2017-10-20abs ↗pdf ↗

FQ-Conv quantizes CNNs for efficient inference with low-precision weights and activations.

problem Reducing precision in DNNs leads to reduced accuracy.
method Fully quantized convolutional neural networks (FQ-Conv) using novel quantization and training techniques.
result Ternary-weight CNNs perform nearly as well as full-precision networks.

High-precision machine learning reduces particle physics simulations by orders of magnitude.

problem Reducing computational burden in particle physics simulations.
method Developed optimal training strategies and tuned machine learning regressors, including Deep Neural Networks with skip connections and boosted decision trees.
result Significantly reduced computational time by factors of 10^3 to 10^6 over first-principles simulations.

With ever-increasing computational demand for deep learning, it is critical to investigate the implications of the numeric representation and precision of DNN model weights and activations on computational efficiency. In this work, we explore unconventional narrow-precision floating-point representations as it relates …

2018-08-07abs ↗pdf ↗

Efficient Bitwidth Search optimizes neural network quantization for better performance.

problem Finding optimal bitwidth for weights and activations of each layer efficiently.
method EBS algorithm reusing meta weights and binary decomposition for efficient mixed precision convolution.
result Mixed precision QNN outperforms uniform bitwidth and other techniques on CIFAR10 and ImageNet.

Deep neural networks have enabled progress in a wide variety of applications. Growing the size of the neural network typically results in improved accuracy. As model sizes grow, the memory and compute requirements for training these models also increases. We introduce a technique to train deep neural networks using hal…

2017-10-10abs ↗pdf ↗

Paper proves existence of a universal codebook for low-precision quantization.

problem Optimizing low-precision approximation of matrix products in machine learning.
method Develops a universal codebook that is near-optimal for all possible statistics of input data.
result Proves existence of a universal codebook with a 0.11 bit per dimension reduction in rate.

It is well-known that the precision of data, hyperparameters, and internal representations employed in learning systems directly impacts its energy, throughput, and latency. The precision requirements for the training algorithm are also important for systems that learn on-the-fly. Prior work has shown that the data and…

2016-07-03abs ↗pdf ↗

New DTI model using self-attention molecule representation outperforms state-of-the-art.

problem Predicting drug-target interactions to reduce costs and improve personalized medicine.
method Proposes a new molecule representation using self-attention and a new DTI model.
result Our DTI model outperforms state-of-the-art by up to 4.9% points in precision-recall.

StatQAT optimizes quantization for deep networks, reducing computational cost and memory usage.

problem Optimal quantization parameters selection for deep neural networks with diverse data distributions.
method Statistical error analysis framework for uniform and floating-point quantization, iterative and analytic quantizers designed for arbitrary and Gaussian-like distributions.
result Improved accuracy and stability in training low-precision neural networks.

A new optimization method reduces memory and compute requirements for deep learning.

problem Memory and compute constraints in second-order stochastic optimizers for deep learning.
method Proposes KrAD, a novel factorization to approximate inverse Fisher matrix without inversion, leading to KrADagrad.
result Improves performance over Shampoo for 32-bit precision and comparable/generalization on real datasets.

The paper improves support recovery in high-dimensional precision matrix estimation using meta learning.

problem Support recovery in high-dimensional precision matrix estimation with reduced sample complexity.
method Pooling samples from different tasks and using an improper 1\ell_1-regularized log-determinant Bregman divergence to estimate a single precision matrix.
result The support of the improperly estimated single precision matrix is equal to the true support union with high probability.

Cheetah framework optimizes DNNs for edge devices using low-precision formats.

problem Reducing DNN model size for edge devices while maintaining accuracy.
method Mixed low-precision hardware and software co-design framework using posit and other formats.
result 16-bit posits outperform 16-bit floating point in training, and [5..8]-bit posits improve inference performance.

A new deep learning method using Boolean logic reduces training and inference energy.

problem High computational and energy costs in deep learning training and inference.
method Introduces Boolean weights and inputs for efficient training using Boolean logic.
result Achieves full-precision accuracy in ImageNet classification and surpasses state-of-the-art results in semantic segmentation.

Model explains how stablecoin runs are influenced by large sales and reserve quality.

problem Understanding and predicting stablecoin runs due to large sales and poor reserve quality.
method Global game model addressing both large sales and poor reserve quality, analyzing risk components.
result The probability of a run increases with large sales and decreases with precise public knowledge, but increases with precise private signals when fundamentals are weak.

This work studies scaling laws for low-precision training in high-dimensional linear regression.

problem Optimizing trade-off between model quality and training costs in high-dimensional linear regression.
method Theoretical study of scaling laws for low-precision training within a high-dimensional sketched linear regression framework, analyzing multiplicative and additive quantization.
result Multiplicative quantization maintains full-precision model size, while additive quantization reduces effective model size.

SPEQ improves quantized neural networks by stochastic precision sharing and cosine similarity loss.

problem Improving quantized deep neural networks for edge devices.
method SPEQ combines stochastic precision sharing and cosine similarity loss for knowledge distillation.
result SPEQ outperforms existing methods in various tasks.

Distributed Lion optimizes large model training by reducing communication costs.

problem Training large AI models efficiently with reduced communication costs.
method Adapted Lion optimizer for distributed training, using binary or lower-precision vectors for communication.
result Distributed Lion achieves comparable performance to standard optimizers but with significantly reduced communication bandwidth.

We show that reducible braids which are, in a Garside-theoretical sense, as simple as possible within their conjugacy class, are also as simple as possible in a geometric sense. More precisely, if a braid belongs to a certain subset of its conjugacy class which we call the stabilized set of sliding circuits, and if it …

2010-08-02abs ↗pdf ↗

Develops a method for constructing KBs with tunable precision for subjective and factual attributes.

problem Complexity in measuring subjective attributes complicates precision estimation in KBs.
method Probabilistically models user consensus with respect to each entity-attribute pair, using neural networks to fit the model.
result Learned models can successfully control KB's precision and outperform baselines in attribute prediction.

SCOPE estimator improves covariance and precision matrix estimation.

problem Estimating covariance and precision matrices accurately.
method Distributionally robust optimization with convex spectral divergence.
result SCOPE estimator reduces spectral bias and improves condition number.

This paper automates the creation of low precision deep learning operators for mobile devices.

problem Deploying large deep learning models on low power devices is challenging due to limited compute capabilities and energy budgets.
method Introduces a workflow to generate high-performance low precision deep learning operators for multiple CPU architectures, including optimizations like memory tiling and vectorization.
result 1-bit and 2-bit convolutions achieve up to 16x and 2.3x speedups over 16-bit integer baselines on ARM Cortex-A53 CPU.

The study analyzes convergence of adaptive optimizers under low-precision training.

problem Understanding why low-precision training remains effective for large models.
method Developed a theoretical framework for analyzing convergence of adaptive optimizers under floating-point quantization.
result Adaptive optimizers retain convergence rates close to full-precision methods under logarithmic mantissa scaling.

A new method selects inducing points to optimize high-throughput Bayesian optimisation.

problem Current inducing point selection methods sacrifice high-fidelity modeling of promising regions.
method Information-theoretic criterion to select inducing points maximizing global and maximum value uncertainties.
result Surrogate models support high-precision high-throughput Bayesian optimisation.

We develop Hodge theory for a Riemannian manifold (M,g)(M,g) with a background closed 3-form, H. Precisely, we prove that if the metric connections with torsion ±H\pm H have holonomy groups G±G_\pm, then the dHd^H-Laplacian preserves the irreducible representations of the Lie algebras of the holonomy groups on the space o…

2012-08-17abs ↗pdf ↗

Proposes a new method for selecting regularization parameters in sparse precision matrix estimation.

problem Selecting an appropriate regularization parameter for sparse precision matrix estimation.
method Developed a closed-form matrix-valued regularization parameter based on the sampling distribution of optimality conditions.
result The proposed method achieves comparable estimation accuracy and superior support recovery to cross-validation, with significant runtime improvements.