Machine Learning (ML) in low-data settings remains an underappreciated yet crucial problem. This challenge is pronounced in low-to-middle income countries where access to large datasets is often limited or even absent. Hence, data augmentation methods to increase the sample size of datasets needed for ML are key to unlocking the transformative potential of ML in data-deprived regions and domains. Unfortunately, the limited training set constrains traditional tabular synthetic data generators in their ability to generate a large and diverse augmented dataset needed for ML tasks. To address this technical challenge, we introduce CLLM, which leverages the prior knowledge of Large Language Models (LLMs) for data augmentation in the low-data regime. While diverse, not all the data generated by LLMs will help increase utility for a downstream task, as for any generative model. Consequently, we introduce a principled curation process, leveraging learning dynamics, coupled with confidence and uncertainty metrics, to obtain a high-quality dataset. Empirically, on multiple real-world datasets, we demonstrate the superior performance of LLMs in the low-data regime compared to conventional generators. We further show our curation mechanism improves the downstream performance for all generators, including LLMs. Additionally, we provide insights and understanding into the LLM generation and curation mechanism, shedding light on the features that enable them to output high-quality augmented datasets. CLLM paves the way for wider usage of ML in data scarce domains and regions, by allying the strengths of LLMs with a robust data-centric approach.
相關內容
大語言模型是基于海量文本數據訓練的深度學習模型。它不僅能夠生成自然語言文本，還能夠深入理解文本含義，處理各種自然語言任務，如文本摘要、問答、翻譯等。2023年，大語言模型及其在人工智能領域的應用已成為全球科技研究的熱點，其在規模上的增長尤為引人注目，參數量已從最初的十幾億躍升到如今的一萬億。參數量的提升使得模型能夠更加精細地捕捉人類語言微妙之處，更加深入地理解人類語言的復雜性。在過去的一年里，大語言模型在吸納新知識、分解復雜任務以及圖文對齊等多方面都有顯著提升。隨著技術的不斷成熟，它將不斷拓展其應用范圍，為人類提供更加智能化和個性化的服務，進一步改善人們的生活和生產方式。
Decoding of Low-Density Parity Check (LDPC) codes can be viewed as a special case of XOR-SAT problems, for which low-computational complexity bit-flipping algorithms have been proposed in the literature. However, a performance gap exists between the bit-flipping LDPC decoding algorithms and the benchmark LDPC decoding algorithms, such as the Sum-Product Algorithm (SPA). In this paper, we propose an XOR-SAT solver using log-sum-exponential functions and demonstrate its advantages for LDPC decoding. This is then approximated using the Margin Propagation formulation to attain a low-complexity LDPC decoder. The proposed algorithm uses soft information to decide the bit-flips that maximize the number of parity check constraints satisfied over an optimization function. The proposed solver can achieve results that are within $0.1$dB of the Sum-Product Algorithm for the same number of code iterations. It is also at least 10x lesser than other Gradient-Descent Bit Flipping decoding algorithms, which are also bit-flipping algorithms based on optimization functions. The approximation using the Margin Propagation formulation does not require any multipliers, resulting in significantly lower computational complexity than other soft-decision Bit-Flipping LDPC decoders.
Sound-soft fractal screens can scatter acoustic waves even when they have zero surface measure. To solve such scattering problems we make what appears to be the first application of the boundary element method (BEM) where each BEM basis function is supported in a fractal set, and the integration involved in the formation of the BEM matrix is with respect to a non-integer order Hausdorff measure rather than the usual (Lebesgue) surface measure. Using recent results on function spaces on fractals, we prove convergence of the Galerkin formulation of this ``Hausdorff BEM'' for acoustic scattering in $\mathbb{R}^{n+1}$ ($n=1,2$) when the scatterer, assumed to be a compact subset of $\mathbb{R}^n\times\{0\}$, is a $d$-set for some $d\in (n-1,n]$, so that, in particular, the scatterer has Hausdorff dimension $d$. For a class of fractals that are attractors of iterated function systems, we prove convergence rates for the Hausdorff BEM and superconvergence for smooth antilinear functionals, under certain natural regularity assumptions on the solution of the underlying boundary integral equation. We also propose numerical quadrature routines for the implementation of our Hausdorff BEM, along with a fully discrete convergence analysis, via numerical (Hausdorff measure) integration estimates and inverse estimates on fractals, estimating the discrete condition numbers. Finally, we show numerical experiments that support the sharpness of our theoretical results, and our solution regularity assumptions, including results for scattering in $\mathbb{R}^2$ by Cantor sets, and in $\mathbb{R}^3$ by Cantor dusts.
We present a demonstration of image classification using an echo-state network (ESN) relying on a single simulated spintronic nanostructure known as the vortex-based spin-torque oscillator (STVO) delayed in time. We employ an ultrafast data-driven simulation framework called the data-driven Thiele equation approach (DD-TEA) to simulate the STVO dynamics. This allows us to avoid the challenges associated with repeated experimental manipulation of such a nanostructured system. We showcase the versatility of our solution by successfully applying it to solve classification challenges with the MNIST, EMNIST-letters and Fashion MNIST datasets. Through our simulations, we determine that within an ESN with numerous learnable parameters the results obtained using the STVO dynamics as an activation function are comparable to the ones obtained with other conventional nonlinear activation functions like the reLU and the sigmoid. While achieving state-of-the-art accuracy levels on the MNIST dataset, our model's performance on EMNIST-letters and Fashion MNIST is lower due to the relative simplicity of the system architecture and the increased complexity of the tasks. We expect that the DD-TEA framework will enable the exploration of deeper architectures, ultimately leading to improved classification accuracy.
Text-to-image generation has made remarkable progress with the emergence of diffusion models. However, it is still a difficult task to generate images for street views based on text, mainly because the road topology of street scenes is complex, the traffic status is diverse and the weather condition is various, which makes conventional text-to-image models difficult to deal with. To address these challenges, we propose a novel controllable text-to-image framework, named \textbf{Text2Street}. In the framework, we first introduce the lane-aware road topology generator, which achieves text-to-map generation with the accurate road structure and lane lines armed with the counting adapter, realizing the controllable road topology generation. Then, the position-based object layout generator is proposed to obtain text-to-layout generation through an object-level bounding box diffusion strategy, realizing the controllable traffic object layout generation. Finally, the multiple control image generator is designed to integrate the road topology, object layout and weather description to realize controllable street-view image generation. Extensive experiments show that the proposed approach achieves controllable street-view text-to-image generation and validates the effectiveness of the Text2Street framework for street views.
We consider nonparametric Bayesian inference in a multidimensional diffusion model with reflecting boundary conditions based on discrete high-frequency observations. We prove a general posterior contraction rate theorem in $L^2$-loss, which is applied to Gaussian priors. The resulting posteriors, as well as their posterior means, are shown to converge to the ground truth at the minimax optimal rate over H\"older smoothness classes in any dimension. Of independent interest and as part of our proofs, we show that certain frequentist penalized least squares estimators are also minimax optimal.
As Multimodal Large Language Models (MLLMs) grow in size, adapting them to specialized tasks becomes increasingly challenging due to high computational and memory demands. Indeed, traditional fine-tuning methods are costly, due to the need for extensive, task-specific training. While efficient adaptation methods exist that aim to reduce these costs, in practice they suffer from shallow inter-modal alignment, which severely hurts model effectiveness. To tackle these computational challenges and improve inter-modal alignment, we introduce the MultiWay-Adapter (MWA), a novel framework featuring an 'Alignment Enhancer'. This enhancer deepens inter-modal alignment, enabling high transferability with minimal tuning effort. Our experiments show that unlike prior efficient tuning approaches, MWA maintains model effectiveness, while reducing training time by up-to 57%. MWA is also lightweight, increasing model size by only 2-3% (in terms of parameters) for state-of-the-art foundation models like BEiT-3 Large. These results demonstrate that MWA provides an efficient and effective adaptation method for MLLMs, significantly broadening their applicability.
Speech intelligibility can be degraded due to multiple factors, such as noisy environments, technical difficulties or biological conditions. This work is focused on the development of an automatic non-intrusive system for predicting the speech intelligibility level in this latter case. The main contribution of our research on this topic is the use of Long Short-Term Memory (LSTM) networks with log-mel spectrograms as input features for this purpose. In addition, this LSTM-based system is further enhanced by the incorporation of a simple attention mechanism that is able to determine the more relevant frames to this task. The proposed models are evaluated with the UA-Speech database that contains dysarthric speech with different degrees of severity. Results show that the attention LSTM architecture outperforms both, a reference Support Vector Machine (SVM)-based system with hand-crafted features and a LSTM-based system with Mean-Pooling.
Support Vector Machines (SVMs) are an important tool for performing classification on scattered data, where one usually has to deal with many data points in high-dimensional spaces. We propose solving SVMs in primal form using feature maps based on trigonometric functions or wavelets. In small dimensional settings the Fast Fourier Transform (FFT) and related methods are a powerful tool in order to deal with the considered basis functions. For growing dimensions the classical FFT-based methods become inefficient due to the curse of dimensionality. Therefore, we restrict ourselves to multivariate basis functions, each one of them depends only on a small number of dimensions. This is motivated by the well-known sparsity of effects and recent results regarding the reconstruction of functions from scattered data in terms of truncated analysis of variance (ANOVA) decomposition, which makes the resulting model even interpretable in terms of importance of the features as well as their couplings. The usage of small superposition dimensions has the consequence that the computational effort no longer grows exponentially but only polynomially with respect to the dimension. In order to enforce sparsity regarding the basis coefficients, we use the frequently applied $\ell_2$-norm and, in addition, $\ell_1$-norm regularization. The found classifying function, which is the linear combination of basis functions, and its variance can then be analyzed in terms of the classical ANOVA decomposition of functions. Based on numerical examples we show that we are able to recover the signum of a function that perfectly fits our model assumptions. We obtain better results with $\ell_1$-norm regularization, both in terms of accuracy and clarity of interpretability.
Neurons in the brain communicate information via punctual events called spikes. The timing of spikes is thought to carry rich information, but it is not clear how to leverage this in digital systems. We demonstrate that event-based encoding is efficient for audio compression. To build this event-based representation we use a deep binary auto-encoder, and under high sparsity pressure, the model enters a regime where the binary event matrix is stored more efficiently with sparse matrix storage algorithms. We test this on the large MAESTRO dataset of piano recordings against vector quantized auto-encoders. Not only does our "Spiking Music compression" algorithm achieve a competitive compression/reconstruction trade-off, but selectivity and synchrony between encoded events and piano key strikes emerge without supervision in the sparse regime.
We present ResMLP, an architecture built entirely upon multi-layer perceptrons for image classification. It is a simple residual network that alternates (i) a linear layer in which image patches interact, independently and identically across channels, and (ii) a two-layer feed-forward network in which channels interact independently per patch. When trained with a modern training strategy using heavy data-augmentation and optionally distillation, it attains surprisingly good accuracy/complexity trade-offs on ImageNet. We will share our code based on the Timm library and pre-trained models.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191