HCBM improves deep learning explainability by non-linear concept aggregation.
problem Lack of explainable and accurate predictions in deep learning for high-stake decisions.
method Introduce Hoeffding Concept Bottleneck Models (HCBM) using Hoeffding functional decomposition of gradient-boosted trees for non-linear and sparse concept aggregation.
result HCBM outperforms standard linear CBM and is robust to interconcept leakage.
CBC makes CNNs robust against adversarial attacks with minimal computational overhead.
problem Making CNNs robust against adversarial attacks without increasing computational complexity.
method CBC uses a stacked encoder-convolutional model where an auto-encoder encodes the input image, and the latent representation is used for classification.
result CBC is more robust to adversarial examples and has significantly lower computational complexity.
Improves GAN performance without changing network architecture.
problem Discriminator fails to guide generator effectively.
method Cascading Rejection (CR) module for discriminator.
result Significantly improves GAN performance in terms of FID.
Enhances deep networks robustness with data mollification and label smoothing.
problem Improving deep neural networks' robustness against corruptions.
method Coupling data mollification (image noising and blurring) with label smoothing.
result Improved robustness and uncertainty quantification on corrupted image benchmarks.
Improved image reconstruction and anomaly detection using hierarchical VAEs.
problem VAEs struggle with sharp images and high-level features.
method Added a new branch to hierarchical VAEs to separate high-level and low-level features.
result Results in sharper images and better anomaly detection.
The impact of the maximally possible batch size (for the better runtime) on performance of graphic processing units (GPU) and tensor processing units (TPU) during training and inference phases is investigated. The numerous runs of the selected deep neural network (DNN) were performed on the standard MNIST and Fashion-M…
A new method speeds up spectral normalization for neural nets.
problem Efficiently controlling the spectral norm of convolutional layers.
method Depthwise separable convolutions with spectral normalization.
result Significant reduction in computational and memory costs.
AugMix improves model robustness and uncertainty.
problem Improving model performance under distribution shifts.
method AugMix: a simple data processing technique.
result Significantly improves robustness and uncertainty measures.
Paper presents a GAN model for realistic river image synthesis.
problem Generating high-quality river images for hydrological research.
method Used a Progressive Growing GAN (PGGAN) architecture to overcome training challenges.
result Demonstrated the effectiveness of GANs in generating high-resolution river images.
ERNet reduces CNN bandwidth needs for imaging tasks.
problem High DRAM bandwidth requirements for CNNs in imaging.
method ERNet uses block-based inference and temporary layer expansion.
result ERNet outperforms state-of-the-art models in image denoising and super-resolution.
New method calibrates neural network uncertainty for medical images.
problem Uncalibrated probabilistic outputs from deep neural networks in medical diagnosis.
method Functional space variational inference for Bayesian neural networks.
result Better calibrated uncertainty estimates at lower computational cost.
In this paper we present Percival, a browser-embedded, lightweight, deep learning-powered ad blocker. Percival embeds itself within the browser's image rendering pipeline, which makes it possible to intercept every image obtained during page execution and to perform blocking based on applying machine learning for image…
Feature Quantization improves GAN training stability.
problem Stability issues in GAN training.
method Feature Quantization (FQ) for the discriminator, embedding true and fake data into a shared discrete space.
result FQ-GAN achieves new state-of-the-art performance on various GAN tasks.
Statistical image reconstruction (SIR) methods are studied extensively for X-ray computed tomography (CT) due to the potential of acquiring CT scans with reduced X-ray dose while maintaining image quality. However, the longer reconstruction time of SIR methods hinders their use in X-ray CT in practice. To accelerate st…
Capsule networks improve with dynamic routing using Wasserstein objective.
problem Capsule networks struggle to consistently outperform traditional neural networks.
method Dynamic routing scheme using approximate Wasserstein objective to select capsules.
result Capsule network achieves over 1.2% improvement on CIFAR-10 with fewer parameters.
GACTGAN synthesizes tabular data better with less computational overhead.
problem Synthesizing mixed tabular data while balancing risk and utility.
method Integrates Bayesian posterior approximation with Stochastic Weight Averaging-Gaussian (SWAG) in CTGAN.
result GACTGAN produces better synthetic data with reduced privacy risk.
FedLog reduces communication in federated learning by sharing data summaries.
problem Significant communication overhead in federated learning with large model parameters.
method Shares minimal sufficient statistics via Bayesian inference and differential privacy.
result High learning accuracy with low communication overhead.
DeepDyve uses simpler neural networks to verify DNNs for faults.
problem Faults in DNNs, especially those not crafted by attackers.
method Pre-trained, simpler neural networks for dynamic verification.
result Reduces 90% of risks with 10% overhead.
Deep learning on an edge device requires energy efficient operation due to ever diminishing power budget. Intentional low quality data during the data acquisition for longer battery life, and natural noise from the low cost sensor degrade the quality of target output which hinders adoption of deep learning on an edge d…
A basic question in the theory of fault-tolerant quantum computation is to understand the fundamental resource costs for performing a universal logical set of gates on encoded qubits to arbitrary accuracy. Here we consider qubits encoded with constant space overhead (i.e. finite encoding rate) in the limit of arbitrari…
The performance and efficiency of distributed machine learning (ML) depends significantly on how long it takes for nodes to exchange state changes. Overly-aggressive attempts to reduce communication often sacrifice final model accuracy and necessitate additional ML techniques to compensate for this loss, limiting their…
The augmented Lagrangian (AL) method that solves convex optimization problems with linear constraints has drawn more attention recently in imaging applications due to its decomposable structure for composite cost functions and empirical fast convergence rate under weak conditions. However, for problems such as X-ray co…
A new framework reduces data upload for image classification while protecting user privacy.
problem Data upload limitations and privacy concerns in cloud-based image classification.
method Unsupervised autoencoder training at edge devices, followed by latent vector transmission to server for classifier training.
result The framework reduces communications overhead and protects user data privacy.
Magnetic resonance image (MRI) reconstruction is a severely ill-posed linear inverse task demanding time and resource intensive computations that can substantially trade off {\it accuracy} for {\it speed} in real-time imaging. In addition, state-of-the-art compressed sensing (CS) analytics are not cognizant of the imag…
Distributed optimization is essential for training large models on large datasets. Multiple approaches have been proposed to reduce the communication overhead in distributed training, such as synchronizing only after performing multiple local SGD steps, and decentralized methods (e.g., using gossip algorithms) to decou…
Enhances image quality to improve test-time adaptation accuracy.
problem Reducing accuracy loss due to distribution shift in deep networks.
method Integrates image enhancement with TTA methods to reduce prediction uncertainty.
result TECA method increases accuracy of TTA methods without hyperparameters.
Deep learning reduces training overhead in massive MIMO systems.
problem Reducing training overhead in massive MIMO systems.
method Use of deep learning (NNs) to improve CSI acquisition and feedback processes.
result Significant improvements in performance and reduced complexity.
We consider the problem of learning a binary classifier from n different data sources, among which at most an η fraction are adversarial. The overhead is defined as the ratio between the sample complexity of learning in this setting and that of learning the same hypothesis class on a single data distribution. We pr…
Deep neural network models used for medical image segmentation are large because they are trained with high-resolution three-dimensional (3D) images. Graphics processing units (GPUs) are widely used to accelerate the trainings. However, the memory on a GPU is not large enough to train the models. A popular approach to …
Paper proposes efficient pruning method for neural networks.
problem Compressing deep neural networks for resource-constrained devices.
method Adaptive sparsity loss for budget-aware optimization during training.
result Demonstrated effectiveness on various architectures and datasets.
A new training method for GANs improves image generation quality.
problem Training GANs is challenging and unstable.
method Consistency regularization to stabilize GAN training.
result CR-GAN achieves best FID scores for unconditional image generation.
We study the collaborative PAC learning problem recently proposed in Blum et al.~\cite{BHPQ17}, in which we have k players and they want to learn a target function collaboratively, such that the learned function approximates the target function well on all players' distributions simultaneously. The quality of the col…
DarkneTZ protects edge devices from DNN model leaks using TEE and model partitioning.
problem Privacy risks of pre-trained DNNs on edge devices through membership inference attacks.
method Model partitioning into sensitive and untrusted parts, leveraging TEE.
result DarkneTZ provides reliable model privacy with minimal performance overhead.
Generative models learn better with data-adaptive noise.
problem Learning heavy-tailed distributions in flow-based models.
method Data-adaptive latent noise using 1D quantile functions optimized via Wasserstein distance.
result Flexibility and effectiveness in learning heavy-tailed and compactly supported distributions.
Computed tomography (CT) is widely used in screening, diagnosis, and image-guided therapy for both clinical and research purposes. Since CT involves ionizing radiation, an overarching thrust of related technical research is development of novel methods enabling ultrahigh quality imaging with fine structural details whi…
New method improves ResNet performance without batch normalization.
problem Improving ResNet performance without batch normalization.
method Adapted Weight Standardization to maintain signal propagation.
result Highly performant ResNets achieve state-of-the-art performance on ImageNet.
Adaptive batch sizes improve local gradient methods in distributed training.
problem Communication bottlenecks in distributed deep learning.
method Adaptive batch size strategies for local gradient methods.
result Adaptive batch sizes reduce minibatch gradient variance and improve training efficiency.
Convolutional GANs favor low spatial frequencies, affecting fine detail generation.
problem Understanding GANs' limitations in high spatial frequency learning.
method Proposed method to manipulate GANs' bias against high spatial frequencies.
result Convolutional GANs have a bias against learning high spatial frequencies.
EnVAE uses energy score for likelihood-free VAEs, improving image reconstructions.
problem Likelihood misspecification in VAEs leads to blurry reconstructions and poor data fidelity.
method Deterministic decoder, energy score as reconstruction loss, fast variant for efficiency.
result EnVAE achieves superior reconstruction and generation quality compared to likelihood-based baselines.
SRG improves optimization efficiency with reduced memory and computation overhead.
problem Limited applicability of variance-reduced optimization algorithms.
method SRG (stochastic reweighted gradient) using importance sampling.
result SRG outperforms SGD and provably improves convergence in strongly-convex cases.
Improved satellite image captions enhance descriptiveness without large models.
problem Extracting meaningful text from satellite imagery.
method Evaluated seven models on a large benchmark, extended vocabulary, and introduced a novel confusion matrix.
result Reduced model size by 100x without sacrificing accuracy, offering new deployment opportunities.
Recent advances in conditional recurrent language modelling have mainly focused on network architectures (e.g., attention mechanism), learning algorithms (e.g., scheduled sampling and sequence-level training) and novel applications (e.g., image/video description generation, speech recognition, etc.) On the other hand, …
New method for scalable set encoding with unbiased gradient approximation.
problem Limited expressive power and large set training issues in set functions.
method Universally MBC (UMBC) class of set functions and efficient MBC training algorithm.
result Unbiased approximation of full set gradient with constant memory overhead.
ABNN converts pre-trained DNNs into BNNs for reliable uncertainty quantification.
problem Uncertainty quantification in deep neural networks (DNNs) is challenging and critical for real-world applications.
method Adaptable Bayesian Neural Network (ABNN) that transforms pre-trained DNNs into BNNs with minimal overhead.
result ABNN achieves state-of-the-art performance in image classification and semantic segmentation tasks.
Federated LIDAR aided beam selection reduces mmWave beam search overhead.
problem Efficient link configuration in mmWave communication systems with reduced beam search overhead.
method Federated learning of LIDAR data to train a shared neural network for beam selection.
result Proposed method significantly outperforms previous works in performance and complexity.
MaxUp improves neural network training by minimizing worst case loss.
problem Improving generalization of neural network training.
method Generate augmented data with random perturbations, minimize maximum loss.
result Consistently outperforms existing methods on various tasks.
Two main obstacles preventing the widespread adoption of variational Bayesian neural networks are the high parameter overhead that makes them infeasible on large networks, and the difficulty of implementation, which can be thought of as "programming overhead." MC dropout [Gal and Ghahramani, 2016] is popular because it…
Drone optimizes collaborative learning for neural networks.
problem Training a shared neural network with heterogeneous data from multiple devices.
method Drone collects and aggregates model parameters from devices, optimizing trajectory for best accuracy.
result Significant improvement in final accuracy and speedup in training time.