The Gap between Theory and Practice in Function Approximation with Deep Neural Networks

Recently deep-learning (DL) techniques have been widely adopted in side-channel power analysis. A DL-assisted SCA generally consists of two phases: a deep neural network (DNN) training phase and a follow-on attack phase using the trained DNN. However, currently the two phases are not well aligned, as there is no conclusion on what metric used in the training can result in the most effective attack in the second phase. When traditional loss functions such as negative log-likelihood (NLL) are used in training a DNN, the trained model does not yield optimal follow-on attack. Recently some information theoretical SCA leakage metrics are proposed, either as the validation metric to stop the DNN training with traditional loss functions, or as both the validation metric and the training loss function. None of those proposed metrics, however, directly measures the SCA effectiveness. We propose to conduct DNN training directly with a common SCA effectiveness metric, Guessing Entropy (GE). We overcome the prior practical difficulty of using GE in DNN training by utilizing the GEEA estimation algorithm introduced in CHES 2020. We show that using GEEA as either the validation metric or the loss function produces DNN models that lead to much more effective follow-on attacks. Our work consolidates the DL-assisted SCA framework with a consistent metric, which shows great potential to be adopted as the universal SCA-oriented DNN training framework.

Survey on Deep Neural Networks in Speech and Vision Systems

Deep Learning (DL) has become a prominent machine learning technique due to the availability of efficient computational resources in the form of Graphics Processing Units (GPUs), large-scale datasets and a variety of models. The newer generation of GPUs are being designed with special emphasis on optimizing performance for DL applications. Also, the availability of easy-to-use DL frameworks—like PyTorch and TensorFlow— has enhanced productivity of domain experts to work on their custom DL applications from diverse domains. However, existing Deep Neural Network (DNN) training approaches may not fully utilize the newly emerging powerful GPUs like the NVIDIA A100—this is the primary issue that we address in this paper. Our motivating analyses show that the GPU utilization on NVIDIA A100 can be as low as 43% using traditional DNN training approaches for small-to-medium DL models and input data size. This paper proposes AccDP—a data-parallel distributed DNN training approach—to accelerate GPU-based DL applications. AccDP exploits the Message Passing Interface (MPI) communication library coupled with the NVIDIA’s Multi-Process Service (MPS) to increase the amount of work assigned to parallel GPUs resulting in higher utilization of compute resources. We evaluate our proposed design on different small-to-medium DL models and input sizes on the state-of-the-art HPC clusters. By injecting more parallelism into DNN training using our approach, the evaluation shows up to 58% improvement in training performance on a single GPU and up to 62% on 16 GPUs compared to regular DNN training. Furthermore, we conduct an in-depth characterization to determine the impact of several DNN training factors and best practices—including the batch size and the number of data loading workers— to optimally utilize GPU devices. To the best of our knowledge, this is the first work that explores the use of MPS and MPI to maximize the utilization of GPUs in distributed DNN training.

We present a physics‐informed deep neural network (DNN) method for estimating hydraulic conductivity in saturated and unsaturated flows governed by Darcy's law. For saturated flow, we approximate hydraulic conductivity and head with two DNNs and use Darcy's law in addition to measurements of hydraulic conductivity and head to train these DNNs. For unsaturated flow, we approximate unsaturated conductivity function and capillary pressure with DNNs and train these DNNs using measurements of capillary pressure and the Richards equation. Because it is difficult to measure unsaturated conductivity in the field, we assume that no measurements of unsaturated conductivity are available. The proposed approach enforces the partial differential equation (PDE) (Darcy or Richards equation) constraints by minimizing the PDE residual at select points in the simulation domain. We demonstrate that physics constraints increase the accuracy of DNN approximations of sparsely observed functions and allow for training DNNs when no direct measurements of the functions of interest are available. For the saturated conductivity estimation problem, we show that the physics‐informed DNN method is more accurate than the state‐of‐the‐art maximum a posteriori probability method. For the unsaturated flow in homogeneous porous media, we find that the proposed method can accurately estimate the pressure‐conductivity relationship based on the capillary pressure measurements only, even in the presence of measurement noise.

The sensitivity of wide parameter-space searches for continuous gravitational waves (CWs) is limited by their high computational cost. Deep learning is being studied as an alternative method to replace various aspects of a CW search. In previous work [], new design principles were presented for deep neural network (DNN) search of CWs and such DNNs were trained to perform a targeted search with matched-filtering sensitivity. In this paper, we adapt these design principles to build a DNN architecture for wide parameter-space searches in 10 days of data from two detectors (H1 and L1). We train a DNN for each of the benchmark cases: six all-sky searches and eight directed searches at different frequencies in the search band of 20–1000 Hz. We compare our results to the DNN sensitivity achieved from Dreissigacker and Prix [] and find that our trained DNNs are more sensitive in all the cases. The absolute improvement in detection probability ranges from 6.5% at 20 Hz to 38% at 1000 Hz in the all-sky cases and from 1.5% at 20 Hz to 59.4% at 500 Hz in the directed cases. An all-sky DNN trained on the entire search band of 20–1000 Hz shows a high sensitivity at all frequencies, providing a proof of concept for training a single DNN to perform the entire search. We also study the generalization of the DNN performance to signals with different signal amplitude, frequency, and the dependence of the DNN sensitivity on sky position. Published by the American Physical Society 2024

The rapid advancements and widespread application of deep neural networks (DNNs), coupled with their reliance on sensitive and private data, have sparked growing concerns regarding data privacy and the ''right to be forgotten''. To address these concerns, machine unlearning has been proposed to efficiently eliminate the influence of specific training data from trained DNNs. However, existing machine unlearning methods struggle with the large number of parameters in trained DNNs, which lead to slow execution and high memory consumption, making them impractical for large-scale models. In this paper, we shift our focus to the small set of weights in the final classification layer of DNNs, which are defined as as ''prototypes'' for different classes. Our key observation is that the prototype associated with the unlearned training data undergoes a significant shift, whereas prototypes of unrelated classes exhibit only minor changes when comparing the prototypes of original and retrained models. Based on this observation, we propose a novel machine unlearning approach that efficiently achieves machine unlearning by directly adjusting the prototypes of DNNs. We first introduce Naive Prototype Surgery (Naive PS), a fast and simplified method that uses a closed-form solution to approximate unlearning effect by directly adjusting the prototype associated with the unlearned data. Next, we propose Prototype Surgery (PS), which incorporates soft label information to fine-tune the prototypes of all classes, to achieve a more effective unlearning. Both methods achieve data unlearning by only modifying the prototypes in the DNNs, thus avoiding the challenges posed by the large number of model parameters. Extensive experiments on four datasets demonstrate that our methods significantly accelerate the unlearning process while achieving comparable results to five existing methods in terms of both unlearning performance and privacy guarantee.

Introduction: Patients with non-ischemic DCM are at risk of life-threatening ventricular arrhythmias. Improved patient selection for ICD implantation is warranted. Deep neural networks (DNN) can discover complex ECG features without the need for hand-crafted feature extraction. This study aimed to distinguish patients at risk of arrhythmias and detect and visualize ECG characteristics that attribute to the DNN classification. Methods: We trained a DNN model on baseline raw ECG samples of adult patients with DCM for the composite outcome of sustained ventricular arrhythmia, (aborted) sudden cardiac death and appropriate ICD therapy in the University Medical Centre Utrecht. Gradient-Weighted Class Activation Mapping was used for visual interpretation of the network. DNN performance was validated both internally and in an external cohort from the Maastricht University Medical University Center. Results: 26.628 ECGs were used for pre-training the network. In total, 317 patients with DCM were included in the primary cohort with a median age of 52 years (IQR 42 - 61), median LVEF of 25% (IQR 20-33), NYHA II and III (32% and 25%). 233 patients were ICD recipients. Subsequently 84 patients reached the outcome (median follow-up duration of 3.96 years (IQR 1.94 - 7.00)). The DNN focused on subtle but predictive differences in the terminal QRS-segment, the T-wave, and the P-wave (figure). The AUC of the DNN was 0.70 in cross-validation. External validation was similar, with an AUC of 0.66. Conclusion: In this study we trained and validated a deep neural network for prediction of life-threatening arrhythmic events. Even though the discriminatory changes on ECG segments were very subtle, the trained DNN was able to predict life-threatening events, focusing on the terminal QRS-segment, the T-wave, and the P-wave. Future studies are warranted to elucidate the identified ECG abnormalities and understand their electrophysiological substrates for LTA prediction in DCM.

Deep Neural Networks (DNNs) tend to go deeper and wider, which poses a significant challenge to the training of DNNs, due to the limited memory capacity of DNN accelerators. Existing solutions for memory-efficient DNN training are densely coupled with the application features of DNN workloads, e.g., layer structures or computational graphs of DNNs are necessary for these solutions. This would result in weak versatility for DNNs with sophisticated layer structures or complicated computation graphs. These schemes usually need to be re-implemented or re-adapted due to the new layer structures or the unusual operators in the computational graphs introduced by these DNNs. In this article, we review the memory pressure issues of DNN training from the perspective of runtime systems and model the memory access behaviors of DNN workloads. We identify the iterative, regularity , and extremalization properties of memory access patterns for DNN workloads. Based on these observations, we propose AppObMem, an application-oblivious memory scheduling system. AppObMem automatically traces the memory behaviors of DNN workloads and schedules the memory swapping to reduce the memory pressure of the device accelerators without the perception of high-level information of layer structures or computation graphs. Evaluations on a variety of DNN models show that, AppObMem obtains 40–60% memory savings with acceptable performance loss. AppObMem is also competitive with other open sourced SOTA schemes.

Distributed processing using high-performance computing resources is essential for developers to train large-scale deep neural networks (DNNs). The major impediment to distributed DNN training is the communication bottleneck during the parameter exchange among the distributed DNN training workers. The communication bottleneck increases training time and decreases the utilization of the computational resources. Our previous study, SoftMemoryBox (SMB1) presented considerably superior performance compared to message passing interface (MPI) in the parameter communication of distributed DNN training. However, SMB1 had disadvantages such as the limited scalability of the distributed DNN training due to the restricted communication bandwidth from a single memory server, inability to provide a synchronization function for the shared memory buffer, and low portability/usability as a consequence of the kernel-level implementation. This paper proposes a scalable, shared memory buffer framework, called SoftMemoryBox II (SMB2), which overcomes the shortcomings of SMB1. With SMB2, distributed training processes can easily share virtually unified shared memory buffers composed of memory segments provided from remote memory servers and can exchange DNN parameters at high speed through the shared memory buffer. The scalable communication bandwidth of the SMB2 framework facilitates the reduction of DNN distributed training times compared to SMB1. According to intensive evaluation results, the communication bandwidth of the proposed SMB2 is 6.3 times greater than that of SMB1 when the SMB2 framework is scaled out to use eight memory servers. Moreover, the training time of SMB2-based asynchronous distributed training of five DNN models is up to 2.4 times faster than SMB1-based training.

Over the last decade, the deep neural networks are a hot topic in machine learning. It is breakthrough technology in processing images, video, speech, text and audio. Deep neural network permits us to overcome some limitations of a shallow neural network due to its deep architecture. In this paper we investigate the nature of unsupervised learning in restricted Boltzmann machine. We have proved that maximization of the log-likelihood input data distribution of restricted Boltzmann machine is equivalent to minimizing the cross-entropy and to special case of minimizing the mean squared error. Thus the nature of unsupervised learning is invariant to different training criteria. As a result we propose a new technique called "REBA" for the unsupervised training of deep neural networks. In contrast to Hinton's conventional approach to the learning of restricted Boltzmann machine, which is based on linear nature of training rule, the proposed technique is founded on nonlinear training rule. We have shown that the classical equations for RBM learning are a special case of the proposed technique. As a result the proposed approach is more universal in contrast to the traditional energy-based model. We demonstrate the performance of the REBA technique using wellknown benchmark problem. The main contribution of this paper is a novel view and new understanding of an unsupervised learning in deep neural networks.

In this invited paper, we present deep neural network (DNN) training accelerator designs in both ASIC and FPGA. The accelerators implements stochastic gradient descent based training algorithm in 16-bit fixed-point precision. A new cyclic weight storage and access scheme enables using the same off-the-shelf SRAMs for non-transpose and transpose operations during feed-forward and feed-backward phases, respectively, of the DNN training process. Including the cyclic weight scheme, the overall DNN training processor is implemented in both 65nm CMOS ASIC and Intel Stratix-10 FPGA hardware. We collectively report the ASIC and FPGA training accelerator results.

Distant speech recognition (DSR) remains to be an open challenge, even for the state-of-the-art deep neural network (DNN) models. Previous work has attempted to improve DNNs under constantly distant speech. However, in real applications, the speaker-microphone distance (SMD) can be quite dynamic, varying even within a single utterance. This paper investigates how to alleviate the impact of dynamic SMD on DNN models. Our solution is to incorporate the frame-level SMD information into DNN training. Generation of the SMD information relies on a universal extractor that is learned on a meeting corpus. We study the utility of different architectures in instantiating the SMD extractor. On our target acoustic modeling task, two approaches are proposed to build distance-aware DNN models using the SMD information: simple concatenation and distance adaptive training (DAT). Our experiments show that in the simplest case, incorporating the SMD descriptors improves word error rates of DNNs by 5.6% relative. Further optimizing SMD extraction and integration results in more gains.

The recent surge of interest in Deep Neural Networks (DNNs) has led to increasingly complex networks that tax computational and memory resources. Many DNNs presently use 16-bit or 32-bit floating point operations. Significant performance and power gains can be obtained when DNN accelerators support low-precision numerical formats. Despite considerable research, there is still a knowledge gap on how low-precision operations can be realized for both DNN training and inference. In this work, we propose a DNN architecture, Deep Positron, with posit numerical format operating successfully at ≤8 bits for inference. We propose a precision-adaptable FPGA soft core for exact multiply-and-accumulate for uniform comparison across three numerical formats, fixed, floating-point and posit. Preliminary results demonstrate that 8-bit posit has better accuracy than 8-bit fixed or floating-point for three different low-dimensional datasets. Moreover, the accuracy is comparable to 32-bit floating-point on a Xilinx Virtex-7 FPGA device. The trade-offs between DNN performance and hardware resources, i.e. latency, power, and resource utilization, show that posit outperforms in accuracy and latency at 8-bit and below.

Contact-map of a protein sequence dictates the global topology of structural fold. Accurate prediction of the contact-map is thus essential to protein 3D structure prediction, which is particularly useful for the protein sequences that do not have close homology templates in the Protein Data Bank. We developed a new method, ResPRE, to predict residue-level protein contacts using inverse covariance matrix (or precision matrix) of multiple sequence alignments (MSAs) through deep residual convolutional neural network training. The approach was tested on a set of 158 non-homologous proteins collected from the CASP experiments and achieved an average accuracy of 50.6% in the top-L long-range contact prediction with L being the sequence length, which is 11.7% higher than the best of other state-of-the-art approaches ranging from coevolution coupling analysis to deep neural network training. Detailed data analyses show that the major advantage of ResPRE lies at the utilization of precision matrix that helps rule out transitional noises of contact-maps compared with the previously used covariance matrix. Meanwhile, the residual network with parallel shortcut layer connections increases the learning ability of deep neural network training. It was also found that appropriate collection of MSAs can further improve the accuracy of final contact-map predictions. The standalone package and online server of ResPRE are made freely available, which should bring important impact on protein structure and function modeling studies in particular for the distant- and non-homology protein targets. https://zhanglab.ccmb.med.umich.edu/ResPRE and https://github.com/leeyang/ResPRE. Supplementary data are available at Bioinformatics online.

In recent years, magnetized inductively coupled plasma (MICP) has been proposed as an improved version of inductively coupled plasma to meet the increasing production process requirements. However, due to the more complex structure of the plasma system, numerical simulations face challenges such as modeling difficulty, model convergence issues, and long computation times. In this paper, a deep neural network (DNN) with a multi-hidden layer structure is developed based on deep learning technology to serve as an assisted modeling tool of traditional fluid simulations. This approach aims to study the discharge characteristics and plasma chemistry of argon-oxygen MICP more efficiently. The simulation data from the fluid model is used to train the neural network. The well-trained DNN can efficiently and accurately predict the target plasma characteristics under new discharge parameters, such as electron density, ionization rate, and particle reaction rate. The effectiveness of the DNN is verified by comparing its predictions with fluid simulation results. Compared to the traditional fluid simulation, which takes thousands of seconds, the trained DNN can produce highly consistent prediction results only within 1 s. The computing efficiency is significantly improved. The prediction results of the DNN demonstrate that by adjusting discharge parameters such as magnetic field coil current, power, and O2 ratio, the plasma discharge characteristics and plasma chemistry can be modulated, including charged particle density, ionization rate, chemical reaction rate, and plasma radial uniformity. Additionally, the DNN shows good generalization ability across multiple input and output scales, providing a paradigm for its application in other plasma research fields. This study indicates that the DNN is a feasible method for studying MICP discharge characteristics and offers a novel and effective computational tool for future plasma discharge research based on deep learning methods.