Large Language Models (LLMs) have achieved extensive impacts across various real-world data mining applications. Given the extremely high cost of training or fine-tuning LLMs, parameter-efficient fine-tuning (e.g., LoRA) has emerged as a popular and practical approach for adapting pre-trained general-purpose LLMs to specific downstream tasks. Among the various hyperparameters involved in parameter-efficient fine-tuning of LLMs, the learning rate (LR) plays a crucial role in determining the overall performance. However, it lacks a systematic benchmark framework to explore and understand how different LR policies influence the effectiveness of parameter-efficient LLM fine-tuning, which makes it challenging to select an optimal LR policy. To address this critical research gap, this paper introduces a systematic benchmark, EFT-LR, for assessing and selecting LR policies for effective parameter-efficient fine-tuning of LLMs. We first present a collection of seven popular LR policies spanning three major categories in the literature. We then perform parameter-efficient fine-tuning of LLMs using these LR policies and assess fine-tuned LLMs on eight downstream tasks. Our empirical analysis using EFT-LR provides an in-depth investigation of the impacts of different LR policies on parameter-efficient LLM fine-tuning, offering practical guidelines for practitioners. We provide the source code at https://github.com/mlsysx/EFT-LR.

Language models (LMs) have demonstrated superior performance in detecting fraudulent activities on Blockchains. Nonetheless, the sheer volume of Blockchain data results in excessive memory and computational costs when training LMs from scratch, limiting their capabilities to large-scale applications. In this paper, we present ZipZap, a framework tailored to achieve both parameter and computational efficiency when training LMs on large-scale transaction data. First, with the frequency-aware compression, an LM can be compressed down to a mere 7.5% of its initial size with an imperceptible performance dip. This technique correlates the embedding dimension of an address with its occurrence frequency in the dataset, motivated by the observation that embeddings of low-frequency addresses are insufficiently trained and thus negating the need for a uniformly large dimension for knowledge representation. Second, ZipZap accelerates the speed through the asymmetric training paradigm: It performs transaction dropping and cross-layer parameter-sharing to expedite the pre-training process, while revert to the standard training paradigm for fine-tuning to strike a balance between efficiency and efficacy, motivated by the observation that the optimization goals of pre-training and fine-tuning are inconsistent. Evaluations on real-world, large-scale datasets demonstrate that ZipZap delivers notable parameter and computational efficiency improvements for training LMs. Our implementation is available at: https://github.com/git-disl/ZipZap.

Deep neural network ensembles combine the wisdom of multiple deep neural networks to improve the generalizability and robustness over individual networks. It has gained increasing popularity to study and apply deep ensemble techniques in the deep learning community. Some mission-critical applications utilize a large number of deep neural networks to form deep ensembles to achieve desired accuracy and resilience, which introduces high time and space costs for ensemble execution. However, it still remains a critical challenge whether a small subset of the entire deep ensemble can achieve the same or better generalizability and how to effectively identify these small deep ensembles for improving the space and time efficiency of ensemble execution. This article presents a novel deep ensemble pruning approach, which can efficiently identify smaller deep ensembles and provide higher ensemble accuracy than the entire deep ensemble of a large number of member networks. Our hierarchical ensemble pruning approach (HQ) leverages three novel ensemble pruning techniques. First, we show that the focal ensemble diversity metrics can accurately capture the complementary capacity of the member networks of an ensemble team, which can guide ensemble pruning. Second, we design a focal ensemble diversity based hierarchical pruning approach, which will iteratively find high quality deep ensembles with low cost and high accuracy. Third, we develop a focal diversity consensus method to integrate multiple focal diversity metrics to refine ensemble pruning results, where smaller deep ensembles can be effectively identified to offer high accuracy, high robustness and high ensemble execution efficiency. Evaluated using popular benchmark datasets, we demonstrate that the proposed hierarchical ensemble pruning approach can effectively identify high quality deep ensembles with better classification generalizability while being more time and space efficient in ensemble decision making. We have released the source codes on GitHub at https://github.com/git-disl/HQ-Ensemble .

Deep neural network ensembles hold the potential of improving generalization performance for complex learning tasks. This paper presents formal analysis and empirical evaluation to show that heterogeneous deep ensembles with high ensemble diversity can effectively leverage model learning heterogeneity to boost ensemble robustness. We first show that heterogeneous DNN models trained for solving the same learning problem, e.g., object detection, can significantly strengthen the mean average precision (mAP) through our weighted bounding box ensemble consensus method. Second, we further compose ensembles of heterogeneous models for solving different learning problems, e.g., object detection and semantic segmentation, by introducing the connected component labeling (CCL) based alignment. We show that this two-tier heterogeneity driven ensemble construction method can compose an ensemble team that promotes high ensemble diversity and low negative correlation among member models of the ensemble, strengthening ensemble robustness against both negative examples and adversarial attacks. Third, we provide a formal analysis of the ensemble robustness in terms of negative correlation. Extensive experiments validate the enhanced robustness of heterogeneous ensembles in both benign and adversarial settings. The appendix and source codes are available on GitHub at https://github.com/git-disl/HeteRobust.

Gradient leakage attacks are dominating privacy threats in federated learning, despite the default privacy that training data resides locally at the clients. Differential privacy has been the de facto standard for privacy protection and is deployed in federated learning to mitigate privacy risks. However, much existing literature points out that differential privacy fails to defend against gradient leakage. The paper presents ModelCloak, a principled approach based on differential privacy noise, aiming for safe-sharing client local model updates. The paper is organized into three major components. First, we introduce the gradient leakage robustness trade-off, in search of the best balance between accuracy and leakage prevention. The trade-off relation is developed based on the behavior of gradient leakage attacks throughout the federated training process. Second, we demonstrate that a proper amount of differential privacy noise can offer the best accuracy performance within the privacy requirement under a fixed differential privacy noise setting. Third, we propose dynamic differential privacy noise and show that the privacy-utility trade-off can be further optimized with dynamic model perturbation, ensuring privacy protection, competitive accuracy, and leakage attack prevention simultaneously.

The choice of learning rate (LR) functions and policies has evolved from a simple fixed LR to the decaying LR and the cyclic LR, aiming to improve the accuracy and reduce the training time of Deep Neural Networks (DNNs). This article presents a systematic approach to selecting and composing an LR policy for effective DNN training to meet desired target accuracy and reduce training time within the pre-defined training iterations. It makes three original contributions. First, we develop an LR tuning mechanism for auto-verification of a given LR policy with respect to the desired accuracy goal under the pre-defined training time constraint. Second, we develop an LR policy recommendation system (LRBench) to select and compose good LR policies from the same and/or different LR functions through dynamic tuning, and avoid bad choices, for a given learning task, DNN model, and dataset. Third, we extend LRBench by supporting different DNN optimizers and show the significant mutual impact of different LR policies and different optimizers. Evaluated using popular benchmark datasets and different DNN models (LeNet, CNN3, ResNet), we show that our approach can effectively deliver high DNN test accuracy, outperform the existing recommended default LR policies, and reduce the DNN training time by 1.6-6.7× to meet a targeted model accuracy.

This article introduces a two-phase deep feature engineering framework for efficient learning of semantics enhanced joint embedding, which clearly separates the deep feature engineering in data preprocessing from training the text-image joint embedding model. We use the Recipe1M dataset for the technical description and empirical validation. In preprocessing, we perform deep feature engineering by combining deep feature engineering with semantic context features derived from raw text-image input data. We leverage LSTM to identify key terms, deep NLP models from the BERT family, TextRank, or TF-IDF to produce ranking scores for key terms before generating the vector representation for each key term by using Word2vec. We leverage Wide ResNet50 and Word2vec to extract and encode the image category semantics of food images to help semantic alignment of the learned recipe and image embeddings in the joint latent space. In joint embedding learning, we perform deep feature engineering by optimizing the batch-hard triplet loss function with soft-margin and double negative sampling, taking into account also the category-based alignment loss and discriminator-based alignment loss. Extensive experiments demonstrate that our SEJE approach with deep feature engineering significantly outperforms the state-of-the-art approaches.

This paper presents a transparent Container Network Memory storage device, coined as CNetMem, aiming to address the open problem of unpredictable performance degradation of containers when the working set of an application no longer fits in container memory. First, CNetMem will enable application tenants running in a container to park their working set memory/file to a faster network memory storage by organizing a group of remote memory nodes as remote memory donors. This allows CNetMem to take advantage of remote idle memory on a cluster before resorting to a slow local I/O subsystem like local disk without any modification of host OS or application. Second, CNetMem provides a hybrid batching technique to remove or alleviate performance bottlenecks in the I/O performance critical path for remote memory read/write with replication or disk backup for fault tolerance. Third, CNetMem introduces a rank-based node selection algorithm to find the optimal node for placing remote memory blocks across cluster. This helps CNetMem to reduce the performance impact due to remote memory eviction. Extensive experiments are conducted on three big data applications and four machine learning workloads. The results show that CNetMem achieves up to 172× throughput improvements compared to vanilla Linux and up to 5.9× completion time improvements over existing approaches in big data and ML workload.

Deep neural network ensembles have become attractive learning techniques with better generalizability over individual models. Some mission critical applications may require a large number of deep neural networks to achieve desirable accuracy and generalizability, making the ensemble execution costly with respect to runtime and space. This paper proposes a novel hierarchical ensemble pruning approach, which can effectively examine a given pool of M base models and identify smaller high quality deep ensembles of size $S(\ll M)$ with higher ensemble accuracy than the entire deep ensemble of all M models. Our hierarchical pruning approach, coined as HQ, combines three novel techniques. First, we show that the focal diversity metrics is innovative and can accurately capture the negative correlation among the member models of an ensemble, and the use of focal diversity metrics can boost ensemble accuracy. Second, we introduce a focal-diversity based hierarchical pruning algorithm to progressively identify low-cost ensembles with high ensemble diversity and accuracy. Third, we design a focal diversity consensus method to find smaller deep ensembles with low negative correlation. We demonstrate such ensembles offer high accuracy and high robustness while being more time and space efficient in ensemble decision making. Evaluated using two benchmark datasets, we show that the proposed focal diversity powered hierarchical pruning can find significantly smaller ensembles of deep neural network models while achieving the same or better classification generalizability.

Deep neural networks (DNNs) have demonstrated impressive performance on many challenging machine learning tasks. However, DNNs are vulnerable to adversarial inputs generated by adding maliciously crafted perturbations to the benign inputs. As a growing number of attacks have been reported to generate adversarial inputs of varying sophistication, the defense-attack arms race has been accelerated. In this paper, we present MODEF, a cross-layer model diversity ensemble framework. MODEF intelligently combines unsupervised model denoising ensemble with supervised model verification ensemble by quantifying model diversity, aiming to boost the robustness of the target model against adversarial examples. Evaluated using eleven representative attacks on popular benchmark datasets, we show that MODEF achieves remarkable defense success rates, compared with existing defense methods, and provides a superior capability of repairing adversarial inputs and making correct predictions with high accuracy in the presence of black-box attacks.

Learning Rate (LR) is an important hyper-parameter to tune for effective training of deep neural networks (DNNs). Even for the baseline of a constant learning rate, it is non-trivial to choose a good constant value for training a DNN. Dynamic learning rates involve multi-step tuning of LR values at various stages of the training process and offer high accuracy and fast convergence. However, they are much harder to tune. In this paper, we present a comprehensive study of 13 learning rate functions and their associated LR policies by examining their range parameters, step parameters, and value update parameters. We propose a set of metrics for evaluating and selecting LR policies, including the classification confidence, variance, cost, and robustness, and implement them in LRBench, an LR benchmarking system. LRBench can assist end-users and DNN developers to select good LR policies and avoid bad LR policies for training their DNNs. We tested LRBench on Caffe, an open source deep learning framework, to showcase the tuning optimization of LR policies. Evaluated through extensive experiments, we attempt to demystify the tuning of LR policies by identifying good LR policies with effective LR value ranges and step sizes for LR update schedules.

Big Data has fueled the wide deployment of Deep Learning (DL) in many fields, such as image classification, voice recognition and NLP. The growing number of open source DL software frameworks has put forward high demands on comparative study of their efficiency with respect to both runtime performance and accuracy. This paper presents a brief overview of our empirical evaluation of four representative DL frameworks: TensorFlow, Caffe, Torch and Theano through a comparative analysis and characterization. First, we show that the complex interactions among neural networks (NN), hyper-parameters, their specific runtime implementations and datasets are latent factors for the uncertainty of runtime performance and accuracy. Second, we characterized the CPU/GPU resource usage patterns under different configurations for different frameworks to obtain an in-depth understanding of the impact of different batch sizes. Third, we describe the data loading process of ImageNet for TensorFlow and present an experimental characterization of TensorFlow with respect to its data loading process when the dataset is too large to fit into the main memory of the CPU server. We conjecture that our experimental characterization and analysis can offer empirical guidance for users and application developers to select the right DL frameworks and configurations for their domain-specific learning tasks and datasets.