Building extraction $-$ needed for inventory management and planning of urban environment $-$ is affected by the misalignment between labels and off-nadir source imagery in training data. Teacher-Student learning of noise-tolerant convolutional neural networks (CNNs) is the existing solution, but the Student networks typically have lower accuracy and cannot surpass the Teacher's performance. This paper proposes a supervised domain adaptation (SDA) of encoder-decoder networks (EDNs) between noisy and clean datasets to tackle the problem. EDNs are configured with high-performing lightweight encoders such as EfficientNet, ResNeSt, and MobileViT. The proposed method is compared against the existing Teacher-Student learning methods like knowledge distillation (KD) and deep mutual learning (DML) with three newly developed datasets. The methods are evaluated for different urban buildings (low-rise, mid-rise, high-rise, and skyscrapers), where misalignment increases with the increase in building height and spatial resolution. For a robust experimental design, 43 lightweight CNNs, five optimisers, nine loss functions, and seven EDNs are benchmarked to obtain the best-performing EDN for SDA. The SDA of the best-performing EDN from our study significantly outperformed KD and DML with up to 0.943, 0.868, 0.912, and 0.697 F1 scores in the low-rise, mid-rise, high-rise, and skyscrapers respectively. The proposed method and the experimental findings will be beneficial in training robust CNNs for building extraction.
Self-supervised foundation models have shown great potential in computer vision thanks to the pre-training paradigm of masked autoencoding. Scale is a primary factor influencing the performance of these foundation models. However, these large foundation models often result in high computational cost. This paper focuses on pre-training relatively small vision transformer models that could be efficiently adapted to downstream tasks. Specifically, taking inspiration from knowledge distillation in model compression, we propose a new asymmetric masked distillation (AMD) framework for pre-training relatively small models with autoencoding. The core of AMD is to devise an asymmetric masking strategy, where the teacher model is enabled to see more context information with a lower masking ratio, while the student model is still equipped with a high masking ratio. We design customized multi-layer feature alignment between the teacher encoder and student encoder to regularize the pre-training of student MAE. To demonstrate the effectiveness and versatility of AMD, we apply it to both ImageMAE and VideoMAE for pre-training relatively small ViT models. AMD achieved 84.6% classification accuracy on IN1K using the ViT-B model. And AMD achieves 73.3% classification accuracy using the ViT-B model on the Something-in-Something V2 dataset, a 3.7% improvement over the original ViT-B model from VideoMAE. We also transfer AMD pre-trained models to downstream tasks and obtain consistent performance improvement over the original masked autoencoding. The code and models are available at https://github.com/MCG-NJU/AMD.
Large Language Models (LLMs) have reshaped natural language processing with their impressive capabilities. However, their ever-increasing size has raised concerns about their effective deployment and the need for LLM compression. This study introduces the Divergent Token Metrics (DTMs), a novel approach to assessing compressed LLMs, addressing the limitations of traditional perplexity or accuracy measures that fail to accurately reflect text generation quality. DTMs measure token divergences that allow deeper insights into the subtleties of model compression, in particular, when evaluating components' impacts individually. Utilizing the First Divergent Token Metric (FDTM) in model sparsification reveals that 25% of all attention components can be pruned beyond 90% on the Llama-2 model family, still keeping SOTA performance. For quantization, FDTM suggests that more than 80% of parameters can be naively transformed to int8 without special outlier management. These evaluations indicate the necessity of choosing appropriate compressions for parameters individually -- and that FDTM can identify those -- while standard metrics result in deteriorated outcomes.
Data-Free Knowledge Distillation (DFKD) plays a vital role in compressing the model when original training data is unavailable. Previous works for DFKD in NLP mainly focus on distilling encoder-only structures like BERT on classification tasks, which overlook the notable progress of generative language modeling. In this work, we propose a novel DFKD framework, namely DFKD-T$^{3}$, where the pretrained generative language model can also serve as a controllable data generator for model compression. This novel framework DFKD-T$^{3}$ leads to an end-to-end learnable text-to-text framework to transform the general domain corpus to compression-friendly task data, targeting to improve both the \textit{specificity} and \textit{diversity}. Extensive experiments show that our method can boost the distillation performance in various downstream tasks such as sentiment analysis, linguistic acceptability, and information extraction. Furthermore, we show that the generated texts can be directly used for distilling other language models and outperform the SOTA methods, making our method more appealing in a general DFKD setting. Our code is available at https://gitee.com/mindspore/models/tree/master/research/nlp/DFKD\_T3.
Large-scale pre-trained language models (LLMs) have demonstrated exceptional performance in various natural language processing (NLP) tasks. However, the massive size of these models poses huge challenges for their deployment in real-world applications. While numerous model compression techniques have been proposed, most of them are not well-suited for achieving extreme model compression when there is a significant gap in model scale. In this paper, we introduce a novel compression paradigm called Retrieval-based Knowledge Transfer (RetriKT), which effectively transfers the knowledge of LLMs to extremely small-scale models (e.g., 1%). In particular, our approach extracts knowledge from LLMs to construct a knowledge store, from which the small-scale model can retrieve relevant information and leverage it for effective inference. To improve the quality of the model, soft prompt tuning and Proximal Policy Optimization (PPO) reinforcement learning techniques are employed. Extensive experiments are conducted on low-resource tasks from SuperGLUE and GLUE benchmarks. The results demonstrate that the proposed approach significantly enhances the performance of small-scale models by leveraging the knowledge from LLMs.
Large-scale pretrained models such as LXMERT are becoming popular for learning cross-modal representations on text-image pairs for vision-language tasks. According to the lottery ticket hypothesis, NLP and computer vision models contain smaller subnetworks capable of being trained in isolation to full performance. In this paper, we combine these observations to evaluate whether such trainable subnetworks exist in LXMERT when fine-tuned on the VQA task. In addition, we perform a model size cost-benefit analysis by investigating how much pruning can be done without significant loss in accuracy. Our experiment results demonstrate that LXMERT can be effectively pruned by 40%-60% in size with 3% loss in accuracy.
Knowledge distillation (KD) has been a popular and effective method for model compression. One important assumption of KD is that the original training dataset is always available. However, this is not always the case due to privacy concerns and more. In recent years, "data-free" KD has emerged as a growing research topic which focuses on the scenario of performing KD when no data is provided. Many methods rely on a generator network to synthesize examples for distillation (which can be difficult to train) and can frequently produce images that are visually similar to the original dataset, which raises questions surrounding whether privacy is completely preserved. In this work, we propose a new approach to data-free KD that utilizes unnatural OpenGL images, combined with large amounts of data augmentation and adversarial attacks, to train a student network. We demonstrate that our approach achieves state-of-the-art results for a variety of datasets/networks and is more stable than existing generator-based data-free KD methods. Source code will be available in the future.
When considering a model architecture, there are several ways to reduce its memory footprint. Historically, popular approaches included selecting smaller architectures and creating sparse networks through pruning. More recently, randomized parameter-sharing (RPS) methods have gained traction for model compression at start of training. In this paper, we comprehensively assess the trade-off between memory and accuracy across RPS, pruning techniques, and building smaller models. Our findings demonstrate that RPS, which is both data and model-agnostic, consistently outperforms/matches smaller models and all moderately informed pruning strategies, such as MAG, SNIP, SYNFLOW, and GRASP, across the entire compression range. This advantage becomes particularly pronounced in higher compression scenarios. Notably, even when compared to highly informed pruning techniques like Lottery Ticket Rewinding (LTR), RPS exhibits superior performance in high compression settings. This points out inherent capacity advantage that RPS enjoys over sparse models. Theoretically, we establish RPS as a superior technique in terms of memory-efficient representation when compared to pruning for linear models. This paper argues in favor of paradigm shift towards RPS based models. During our rigorous evaluation of RPS, we identified issues in the state-of-the-art RPS technique ROAST, specifically regarding stability (ROAST's sensitivity to initialization hyperparameters, often leading to divergence) and Pareto-continuity (ROAST's inability to recover the accuracy of the original model at zero compression). We provably address both of these issues. We refer to the modified RPS, which incorporates our improvements, as STABLE-RPS.
Visual Transformers have achieved great success in almost all vision tasks, such as classification, detection, and so on. However, the model complexity and the inference speed of the visual transformers hinder their deployments in industrial products. Various model compression techniques focus on directly compressing the visual transformers into a smaller one while maintaining the model performance, however, the performance drops dramatically when the compression ratio is large. Furthermore, several dynamic network techniques have also been applied to dynamically compress the visual transformers to obtain input-adaptive efficient sub-structures during the inference stage, which can achieve a better trade-off between the compression ratio and the model performance. The upper bound of memory of dynamic models is not reduced in the practical deployment since the whole original visual transformer model and the additional control gating modules should be loaded onto devices together for inference. To alleviate two disadvantages of two categories of methods, we propose to unify the static compression and dynamic compression techniques jointly to obtain an input-adaptive compressed model, which can further better balance the total compression ratios and the model performances. Moreover, in practical deployment, the batch sizes of the training and inference stage are usually different, which will cause the model inference performance to be worse than the model training performance, which is not touched by all previous dynamic network papers. We propose a sub-group gates augmentation technique to solve this performance drop problem. Extensive experiments demonstrate the superiority of our method on various baseline visual transformers such as DeiT, T2T-ViT, and so on.
This study proposes a method based on lightweight convolutional neural networks (CNN) and generative adversarial networks (GAN) for apple ripeness and damage level detection tasks. Initially, a lightweight CNN model is designed by optimizing the model's depth and width, as well as employing advanced model compression techniques, successfully reducing the model's parameter and computational requirements, thus enhancing real-time performance in practical applications. Simultaneously, attention mechanisms are introduced, dynamically adjusting the importance of different feature layers to improve the performance in object detection tasks. To address the issues of sample imbalance and insufficient sample size, GANs are used to generate realistic apple images, expanding the training dataset and enhancing the model's recognition capability when faced with apples of varying ripeness and damage levels. Furthermore, by applying the object detection network for damage location annotation on damaged apples, the accuracy of damage level detection is improved, providing a more precise basis for decision-making. Experimental results show that in apple ripeness grading detection, the proposed model achieves 95.6\%, 93.8\%, 95.0\%, and 56.5 in precision, recall, accuracy, and FPS, respectively. In apple damage level detection, the proposed model reaches 95.3\%, 93.7\%, and 94.5\% in precision, recall, and mAP, respectively. In both tasks, the proposed method outperforms other mainstream models, demonstrating the excellent performance and high practical value of the proposed method in apple ripeness and damage level detection tasks.
Sliding Window Sum algorithms have been successfully used for training and inference of Deep Neural Networks. We have shown before how both pooling and convolution 1-D primitives could be expressed as sliding sums and evaluated by the compute kernels with a shared structure. In this paper, we present an extensive study of the Sliding Window convolution technique as a more efficient alternative to the commonly used General Matrix Multiplication (GEMM) based convolution in Deep Neural Networks (DNNs). The Sliding Window technique addresses the memory bloating problem and demonstrates a significant speedup in 2-D convolution. We explore the performance of this technique on a range of implementations, including custom kernels for specific filter sizes. Our results suggest that the Sliding Window computation kernels can outperform GEMM-based convolution on a CPU and even on dedicated hardware accelerators. This could promote a wider adoption of AI on low-power and low-memory devices without the need for specialized hardware. We also discuss the compatibility of model compression methods and optimized network architectures with the Sliding Window technique, encouraging further research in these areas.
Sequential maximization of expected improvement (EI) is one of the most widely used policies in Bayesian optimization because of its simplicity and ability to handle noisy observations. In particular, the improvement function often uses the best posterior mean as the best incumbent in noisy settings. However, the uncertainty associated with the incumbent solution is often neglected in many analytic EI-type methods: a closed-form acquisition function is derived in the noise-free setting, but then applied to the setting with noisy observations. To address this limitation, we propose a modification of EI that corrects its closed-form expression by incorporating the covariance information provided by the Gaussian Process (GP) model. This acquisition function specializes to the classical noise-free result, and we argue should replace that formula in Bayesian optimization software packages, tutorials, and textbooks. This enhanced acquisition provides good generality for noisy and noiseless settings. We show that our method achieves a sublinear convergence rate on the cumulative regret bound under heteroscedastic observation noise. Our empirical results demonstrate that our proposed acquisition function can outperform EI in the presence of noisy observations on benchmark functions for black-box optimization, as well as on parameter search for neural network model compression.
On-device machine learning (ML) promises to improve the privacy, responsiveness, and proliferation of new, intelligent user experiences by moving ML computation onto everyday personal devices. However, today's large ML models must be drastically compressed to run efficiently on-device, a hurtle that requires deep, yet currently niche expertise. To engage the broader human-centered ML community in on-device ML experiences, we present the results from an interview study with 30 experts at Apple that specialize in producing efficient models. We compile tacit knowledge that experts have developed through practical experience with model compression across different hardware platforms. Our findings offer pragmatic considerations missing from prior work, covering the design process, trade-offs, and technical strategies that go into creating efficient models. Finally, we distill design recommendations for tooling to help ease the difficulty of this work and bring on-device ML into to more widespread practice.
The success of modern neural networks has prompted study of the connection between memorisation and generalisation: overparameterised models generalise well, despite being able to perfectly fit (memorise) completely random labels. To carefully study this issue, Feldman proposed a metric to quantify the degree of memorisation of individual training examples, and empirically computed the corresponding memorisation profile of a ResNet on image classification bench-marks. While an exciting first glimpse into what real-world models memorise, this leaves open a fundamental question: do larger neural models memorise more? We present a comprehensive empirical analysis of this question on image classification benchmarks. We find that training examples exhibit an unexpectedly diverse set of memorisation trajectories across model sizes: most samples experience decreased memorisation under larger models, while the rest exhibit cap-shaped or increasing memorisation. We show that various proxies for the Feldman memorization score fail to capture these fundamental trends. Lastly, we find that knowledge distillation, an effective and popular model compression technique, tends to inhibit memorisation, while also improving generalisation. Specifically, memorisation is mostly inhibited on examples with increasing memorisation trajectories, thus pointing at how distillation improves generalisation.
Quantization has emerged as a promising direction for model compression. Recently, data-free quantization has been widely studied as a promising method to avoid privacy concerns, which synthesizes images as an alternative to real training data. Existing methods use classification loss to ensure the reliability of the synthesized images. Unfortunately, even if these images are well-classified by the pre-trained model, they still suffer from low semantics and homogenization issues. Intuitively, these low-semantic images are sensitive to perturbations, and the pre-trained model tends to have inconsistent output when the generator synthesizes an image with poor semantics. To this end, we propose Robustness-Guided Image Synthesis (RIS), a simple but effective method to enrich the semantics of synthetic images and improve image diversity, further boosting the performance of downstream data-free compression tasks. Concretely, we first introduce perturbations on input and model weight, then define the inconsistency metrics at feature and prediction levels before and after perturbations. On the basis of inconsistency on two levels, we design a robustness optimization objective to enhance the semantics of synthetic images. Moreover, we also make our approach diversity-aware by forcing the generator to synthesize images with small correlations in the label space. With RIS, we achieve state-of-the-art performance for various settings on data-free quantization and can be extended to other data-free compression tasks.
Sparse deep learning has become a popular technique for improving the performance of deep neural networks in areas such as uncertainty quantification, variable selection, and large-scale network compression. However, most existing research has focused on problems where the observations are independent and identically distributed (i.i.d.), and there has been little work on the problems where the observations are dependent, such as time series data and sequential data in natural language processing. This paper aims to address this gap by studying the theory for sparse deep learning with dependent data. We show that sparse recurrent neural networks (RNNs) can be consistently estimated, and their predictions are asymptotically normally distributed under appropriate assumptions, enabling the prediction uncertainty to be correctly quantified. Our numerical results show that sparse deep learning outperforms state-of-the-art methods, such as conformal predictions, in prediction uncertainty quantification for time series data. Furthermore, our results indicate that the proposed method can consistently identify the autoregressive order for time series data and outperform existing methods in large-scale model compression. Our proposed method has important practical implications in fields such as finance, healthcare, and energy, where both accurate point estimates and prediction uncertainty quantification are of concern.
Large Vision-Language Models (LVLMs) can understand the world comprehensively by integrating rich information from different modalities, achieving remarkable advancements on various multimodal downstream tasks. However, deploying LVLMs is often problematic due to their massive computational/energy costs and carbon consumption. Such issues make it infeasible to adopt conventional iterative global pruning, which is costly due to computing the Hessian matrix of the entire large model for sparsification. Alternatively, several studies have recently proposed layer-wise pruning approaches to avoid the expensive computation of global pruning and efficiently compress model weights according to their importance within a layer. However, they often suffer from suboptimal model compression due to their lack of a global perspective. To address this limitation in recent efficient pruning methods for large models, we propose Efficient Coarse-to-Fine LayerWise Pruning (ECoFLaP), a two-stage coarse-to-fine weight pruning approach for LVLMs. We first determine the sparsity ratios of different layers or blocks by leveraging the global importance score, which is efficiently computed based on the zeroth-order approximation of the global model gradients. Then, the model performs local layer-wise unstructured weight pruning based on globally-informed sparsity ratios. We validate our proposed method across various multimodal and unimodal models and datasets, demonstrating significant performance improvements over prevalent pruning techniques in the high-sparsity regime.
Large Language Models (LLMs) have the ability to solve a variety of tasks, such as text summarization and mathematical questions, just out of the box, but they are often trained with a single task in mind. Due to high computational costs, the current trend is to use prompt instruction tuning to better adjust monolithic, pretrained LLMs for new -- but often individual -- downstream tasks. Thus, how one would expand prompt tuning to handle -- concomitantly -- heterogeneous tasks and data distributions is a widely open question. To address this gap, we suggest the use of \emph{Mixture of Prompts}, or MoPs, associated with smart gating functionality: the latter -- whose design is one of the contributions of this paper -- can identify relevant skills embedded in different groups of prompts and dynamically assign combined experts (i.e., collection of prompts), based on the target task. Additionally, MoPs are empirically agnostic to any model compression technique applied -- for efficiency reasons -- as well as instruction data source and task composition. In practice, MoPs can simultaneously mitigate prompt training "interference" in multi-task, multi-source scenarios (e.g., task and data heterogeneity across sources), as well as possible implications from model approximations. As a highlight, MoPs manage to decrease final perplexity from $\sim20\%$ up to $\sim70\%$, as compared to baselines, in the federated scenario, and from $\sim 3\%$ up to $\sim30\%$ in the centralized scenario.
Privacy-Preserving ML (PPML) based on Homomorphic Encryption (HE) is a promising foundational privacy technology. Making it more practical requires lowering its computational cost, especially, in handling modern large deep neural networks. Model compression via pruning is highly effective in conventional plaintext ML but cannot be effectively applied to HE-PPML as is. We propose Artemis, a highly effective DNN pruning technique for HE-based inference. We judiciously investigate two HE-aware pruning strategies (positional and diagonal) to reduce the number of Rotation operations, which dominate compute time in HE convolution. We find that Pareto-optimal solutions are based fully on diagonal pruning. Artemis' benefits come from coupling DNN training, driven by a novel group Lasso regularization objective, with pruning to maximize HE-specific cost reduction (dominated by the Rotation operations). We show that Artemis improves on prior HE-oriented pruning and can achieve a 1.2-6x improvement when targeting modern convolutional models (ResNet18 and ResNet18) across three datasets.
Federated learning (FL) offers privacy-preserving decentralized machine learning, optimizing models at edge clients without sharing private data. Simultaneously, foundation models (FMs) have gained traction in the artificial intelligence (AI) community due to their exceptional performance across various tasks. However, integrating FMs into FL presents challenges, primarily due to their substantial size and intensive resource requirements. This is especially true when considering the resource heterogeneity in edge FL systems. We present an adaptive framework for Resource-aware Federated Foundation Models (RaFFM) to address these challenges. RaFFM introduces specialized model compression algorithms tailored for FL scenarios, such as salient parameter prioritization and high-performance subnetwork extraction. These algorithms enable dynamic scaling of given transformer-based FMs to fit heterogeneous resource constraints at the network edge during both FL's optimization and deployment stages. Experimental results demonstrate that RaFFM shows significant superiority in resource utilization efficiency and uses fewer resources to deploy FMs to FL. Despite the lower resource consumption, target models optimized by RaFFM achieve performance on par with traditional FL methods applied to full-sized FMs. This is evident across tasks in both natural language processing and computer vision domains.