Learning from limited data has been extensively studied in machine learning, considering that deep neural networks achieve optimal performance when trained using a large amount of samples. Although various strategies have been proposed for centralized training, the topic of federated learning with small datasets remains largely unexplored. Moreover, in realistic scenarios, such as settings where medical institutions are involved, the number of participating clients is also constrained. In this work, we propose a novel federated learning framework, named RepTreeFL. At the core of the solution is the concept of a replica, where we replicate each participating client by copying its model architecture and perturbing its local data distribution. Our approach enables learning from limited data and a small number of clients by aggregating a larger number of models with diverse data distributions. Furthermore, we leverage the hierarchical structure of the client network (both original and virtual), alongside the model diversity across replicas, and introduce a diversity-based tree aggregation, where replicas are combined in a tree-like manner and the aggregation weights are dynamically updated based on the model discrepancy. We evaluated our method on two tasks and two types of data, graph generation and image classification (binary and multi-class), with both homogeneous and heterogeneous model architectures. Experimental results demonstrate the effectiveness and outperformance of RepTreeFL in settings where both data and clients are limited. Our code is available at https://github.com/basiralab/RepTreeFL.
Humans excel at efficiently navigating through crowds without collision by focusing on specific visual regions relevant to navigation. However, most robotic visual navigation methods rely on deep learning models pre-trained on vision tasks, which prioritize salient objects -- not necessarily relevant to navigation and potentially misleading. Alternative approaches train specialized navigation models from scratch, requiring significant computation. On the other hand, self-supervised learning has revolutionized computer vision and natural language processing, but its application to robotic navigation remains underexplored due to the difficulty of defining effective self-supervision signals. Motivated by these observations, in this work, we propose a Self-Supervised Vision-Action Model for Visual Navigation Pre-Training (VANP). Instead of detecting salient objects that are beneficial for tasks such as classification or detection, VANP learns to focus only on specific visual regions that are relevant to the navigation task. To achieve this, VANP uses a history of visual observations, future actions, and a goal image for self-supervision, and embeds them using two small Transformer Encoders. Then, VANP maximizes the information between the embeddings by using a mutual information maximization objective function. We demonstrate that most VANP-extracted features match with human navigation intuition. VANP achieves comparable performance as models learned end-to-end with half the training time and models trained on a large-scale, fully supervised dataset, i.e., ImageNet, with only 0.08% data.
Semantic segmentation plays a crucial role in enabling comprehensive scene understanding for robotic systems. However, generating annotations is challenging, requiring labels for every pixel in an image. In scenarios like autonomous driving, there's a need to progressively incorporate new classes as the operating environment of the deployed agent becomes more complex. For enhanced annotation efficiency, ideally, only pixels belonging to new classes would be annotated. This approach is known as Continual Semantic Segmentation (CSS). Besides the common problem of classical catastrophic forgetting in the continual learning setting, CSS suffers from the inherent ambiguity of the background, a phenomenon we refer to as the "background shift'', since pixels labeled as background could correspond to future classes (forward background shift) or previous classes (backward background shift). As a result, continual learning approaches tend to fail. This paper proposes a Backward Background Shift Detector (BACS) to detect previously observed classes based on their distance in the latent space from the foreground centroids of previous steps. Moreover, we propose a modified version of the cross-entropy loss function, incorporating the BACS detector to down-weight background pixels associated with formerly observed classes. To combat catastrophic forgetting, we employ masked feature distillation alongside dark experience replay. Additionally, our approach includes a transformer decoder capable of adjusting to new classes without necessitating an additional classification head. We validate BACS's superior performance over existing state-of-the-art methods on standard CSS benchmarks.
High-efficiency deep learning (DL) models are necessary not only to facilitate their use in devices with limited resources but also to improve resources required for training. Convolutional neural networks (ConvNets) typically exert severe demands on local device resources and this conventionally limits their adoption within mobile and embedded platforms. This brief presents work toward utilizing static convolutional filters generated from the space of local binary patterns (LBPs) and Haar features to design efficient ConvNet architectures. These are referred to as Structured Ternary Patterns (STePs) and can be generated during network initialization in a systematic way instead of having learnable weight parameters thus reducing the total weight updates. The ternary values require significantly less storage and with the appropriate low-level implementation, can also lead to inference improvements. The proposed approach is validated using four image classification datasets, demonstrating that common network backbones can be made more efficient and provide competitive results. It is also demonstrated that it is possible to generate completely custom STeP-based networks that provide good trade-offs for on-device applications such as unmanned aerial vehicle (UAV)-based aerial vehicle detection. The experimental results show that the proposed method maintains high detection accuracy while reducing the trainable parameters by 40-80%. This work motivates further research toward good priors for non-learnable weights that can make DL architectures more efficient without having to alter the network during or after training.
생성형 AI 이미지 탐지기들이 실세계 조건에서 적대적 예제(adversarial examples)에 심각하게 취약함을 4가지 탐지 방법과 5가지 공격 알고리즘으로 검증하며, 블랙박스 공격만으로도 평균 AUROC를 87.5%에서 약 0%까지 낮출 수 있음을 보였다. 소셜 미디어 업로드 시 발생하는 압축·리사이즈 등의 품질 저하 후에도 공격이 유효하며, 상용 탐지 도구 HIVE에도 성공적으로 적용되었고, RobustCLIP 기반 방어가 강건성을 부분적으로 향상시키나 완전한 해결책에는 미치지 못함을 확인했다.
Novelty detection is the process of identifying the observation(s) that differ in some respect from the training observations (the target class). In reality, the novelty class is often absent during training, poorly sampled or not well defined. Therefore, one-class classifiers can efficiently model such problems. However, due to the unavailability of data from the novelty class, training an end-to-end deep network is a cumbersome task. In this paper, inspired by the success of generative adversarial networks for training deep models in unsupervised and semi-supervised settings, we propose an end-to-end architecture for one-class classification. Our architecture is composed of two deep networks, each of which trained by competing with each other while collaborating to understand the underlying concept in the target class, and then classify the testing samples. One network works as the novelty detector, while the other supports it by enhancing the inlier samples and distorting the outliers. The intuition is that the separability of the enhanced inliers and distorted outliers is much better than deciding on the original samples. The proposed framework applies to different related applications of anomaly and outlier detection in images and videos. The results on MNIST and Caltech-256 image datasets, along with the challenging UCSD Ped2 dataset for video anomaly detection illustrate that our proposed method learns the target class effectively and is superior to the baseline and state-of-the-art methods.
We address the Unsupervised Domain Adaptation (UDA) problem in image classification from a new perspective. In contrast to most existing works which either align the data distributions or learn domain-invariant features, we directly learn a unified classifier for both domains within a high-dimensional homogeneous feature space without explicit domain adaptation. To this end, we employ the effective Selective Pseudo-Labelling (SPL) techniques to take advantage of the unlabelled samples in the target domain. Surprisingly, data distribution discrepancy across the source and target domains can be well handled by a computationally simple classifier (e.g., a shallow Multi-Layer Perceptron) trained in the original feature space. Besides, we propose a novel generative model norm-VAE to generate synthetic features for the target domain as a data augmentation strategy to enhance classifier training. Experimental results on several benchmark datasets demonstrate the pseudo-labelling strategy itself can lead to comparable performance to many state-of-the-art methods whilst the use of norm-VAE for feature augmentation can further improve the performance in most cases. As a result, our proposed methods (i.e. naive-SPL and norm-VAE-SPL) can achieve new state-of-the-art performance with the average accuracy of 93.4% and 90.4% on Office-Caltech and ImageCLEF-DA datasets, and comparable performance on Digits, Office31 and Office-Home datasets with the average accuracy of 97.2%, 87.6% and 67.9% respectively.
Recently, it has been demonstrated that the performance of a deep convolutional neural network can be effectively improved by embedding an attention module into it. In this work, a novel lightweight and effective attention method named Pyramid Squeeze Attention (PSA) module is proposed. By replacing the 3x3 convolution with the PSA module in the bottleneck blocks of the ResNet, a novel representational block named Efficient Pyramid Squeeze Attention (EPSA) is obtained. The EPSA block can be easily added as a plug-and-play component into a well-established backbone network, and significant improvements on model performance can be achieved. Hence, a simple and efficient backbone architecture named EPSANet is developed in this work by stacking these ResNet-style EPSA blocks. Correspondingly, a stronger multi-scale representation ability can be offered by the proposed EPSANet for various computer vision tasks including but not limited to, image classification, object detection, instance segmentation, etc. Without bells and whistles, the performance of the proposed EPSANet outperforms most of the state-of-the-art channel attention methods. As compared to the SENet-50, the Top-1 accuracy is improved by 1.93% on ImageNet dataset, a larger margin of +2.7 box AP for object detection and an improvement of +1.7 mask AP for instance segmentation by using the Mask-RCNN on MS-COCO dataset are obtained. Our source code is available at:https://github.com/murufeng/EPSANet.
There has been significant research done on developing methods for improving robustness to distributional shift and uncertainty estimation. In contrast, only limited work has examined developing standard datasets and benchmarks for assessing these approaches. Additionally, most work on uncertainty estimation and robustness has developed new techniques based on small-scale regression or image classification tasks. However, many tasks of practical interest have different modalities, such as tabular data, audio, text, or sensor data, which offer significant challenges involving regression and discrete or continuous structured prediction. Thus, given the current state of the field, a standardized large-scale dataset of tasks across a range of modalities affected by distributional shifts is necessary. This will enable researchers to meaningfully evaluate the plethora of recently developed uncertainty quantification methods, as well as assessment criteria and state-of-the-art baselines. In this work, we propose the Shifts Dataset for evaluation of uncertainty estimates and robustness to distributional shift. The dataset, which has been collected from industrial sources and services, is composed of three tasks, with each corresponding to a particular data modality: tabular weather prediction, machine translation, and self-driving car (SDC) vehicle motion prediction. All of these data modalities and tasks are affected by real, "in-the-wild" distributional shifts and pose interesting challenges with respect to uncertainty estimation. In this work we provide a description of the dataset and baseline results for all tasks.
Modern machine learning models are complicated. Most of them rely on convoluted latent representations of their input to issue a prediction. To achieve greater transparency than a black-box that connects inputs to predictions, it is necessary to gain a deeper understanding of these latent representations. To that aim, we propose SimplEx: a user-centred method that provides example-based explanations with reference to a freely selected set of examples, called the corpus. SimplEx uses the corpus to improve the user's understanding of the latent space with post-hoc explanations answering two questions: (1) Which corpus examples explain the prediction issued for a given test example? (2) What features of these corpus examples are relevant for the model to relate them to the test example? SimplEx provides an answer by reconstructing the test latent representation as a mixture of corpus latent representations. Further, we propose a novel approach, the Integrated Jacobian, that allows SimplEx to make explicit the contribution of each corpus feature in the mixture. Through experiments on tasks ranging from mortality prediction to image classification, we demonstrate that these decompositions are robust and accurate. With illustrative use cases in medicine, we show that SimplEx empowers the user by highlighting relevant patterns in the corpus that explain model representations. Moreover, we demonstrate how the freedom in choosing the corpus allows the user to have personalized explanations in terms of examples that are meaningful for them.
Though the state-of-the architectures for semantic segmentation, such as HRNet, demonstrate impressive accuracy, the complexity arising from their salient design choices hinders a range of model acceleration tools, and further they make use of operations that are inefficient on current hardware. This paper demonstrates that a simple encoder-decoder architecture with a ResNet-like backbone and a small multi-scale head, performs on-par or better than complex semantic segmentation architectures such as HRNet, FANet and DDRNets. Naively applying deep backbones designed for Image Classification to the task of Semantic Segmentation leads to sub-par results, owing to a much smaller effective receptive field of these backbones. Implicit among the various design choices put forth in works like HRNet, DDRNet, and FANet are networks with a large effective receptive field. It is natural to ask if a simple encoder-decoder architecture would compare favorably if comprised of backbones that have a larger effective receptive field, though without the use of inefficient operations like dilated convolutions. We show that with minor and inexpensive modifications to ResNets, enlarging the receptive field, very simple and competitive baselines can be created for Semantic Segmentation. We present a family of such simple architectures for desktop as well as mobile targets, which match or exceed the performance of complex models on the Cityscapes dataset. We hope that our work provides simple yet effective baselines for practitioners to develop efficient semantic segmentation models.
It is widely believed that the success of deep networks lies in their ability to learn a meaningful representation of the features of the data. Yet, understanding when and how this feature learning improves performance remains a challenge: for example, it is beneficial for modern architectures trained to classify images, whereas it is detrimental for fully-connected networks trained for the same task on the same data. Here we propose an explanation for this puzzle, by showing that feature learning can perform worse than lazy training (via random feature kernel or the NTK) as the former can lead to a sparser neural representation. Although sparsity is known to be essential for learning anisotropic data, it is detrimental when the target function is constant or smooth along certain directions of input space. We illustrate this phenomenon in two settings: (i) regression of Gaussian random functions on the d-dimensional unit sphere and (ii) classification of benchmark datasets of images. For (i), we compute the scaling of the generalization error with number of training points, and show that methods that do not learn features generalize better, even when the dimension of the input space is large. For (ii), we show empirically that learning features can indeed lead to sparse and thereby less smooth representations of the image predictors. This fact is plausibly responsible for deteriorating the performance, which is known to be correlated with smoothness along diffeomorphisms.
The ability to quickly learn a new task with minimal instruction - known as few-shot learning - is a central aspect of intelligent agents. Classical few-shot benchmarks make use of few-shot samples from a single modality, but such samples may not be sufficient to characterize an entire concept class. In contrast, humans use cross-modal information to learn new concepts efficiently. In this work, we demonstrate that one can indeed build a better ${\bf visual}$ dog classifier by ${\bf read}$ing about dogs and ${\bf listen}$ing to them bark. To do so, we exploit the fact that recent multimodal foundation models such as CLIP learn cross-modal encoders that map different modalities to the same representation space. Specifically, we propose a simple strategy for ${\bf cross-modal}$ ${\bf adaptation}$: we treat examples from different modalities as additional few-shot examples. For example, by simply repurposing class names as an additional training sample, we trivially turn any n-shot learning problem into a (n+1)-shot problem. This allows us to produce SOTA results with embarrassingly simple linear classifiers. We show that our approach can be combined with existing methods such as prefix tuning, adapters, and classifier ensembling. Finally, to explore other modalities beyond vision and language, we construct the first (to our knowledge) audiovisual few-shot benchmark and use cross-modal training to improve the performance of both image and audio classification.
DP-SGD와 같은 머신러닝(ML) 알고리즘에 대한 차분 프라이버시(DP) 보장을 보고하는 현재의 관행은 불완전하고 잠재적으로 오해를 초래할 수 있는 그림을 제시합니다. 예를 들어, 어떤 메커니즘에 대해 단일 $(\varepsilon, \delta)$만 알려진 경우, 표준 분석은 학습 데이터 레코드에 대한 매우 정확한 추론 공격이 존재할 수 있음을 보여주지만, 보다 신중한 분석을 통해 살펴보면 대부분의 실용적인 메커니즘에서는 그러한 정확한 공격이 존재하지 않습니다. 이 포지션 페이퍼에서 우리는 ML에서 DP 보장을 전달하는 주요 수단으로 _비점근적_ 가우시안 차분 프라이버시(GDP)를 사용하는 것이 이러한 잠재적 단점을 피할 수 있다고 주장합니다. DP 문헌의 두 가지 최근 발전, 즉 (i) DP-SGD의 프라이버시 프로파일과 $f$-DP 곡선을 임의의 정확도로 계산할 수 있는 오픈소스 수치 어카운턴트, (ii) DP 표현에 대한 의사결정 이론적 메트릭을 활용하여, 우리는 수치 어카운턴트를 사용해 GDP에 대한 비점근적 경계를 제공하는 방법을 보이고, GDP가 해당 메트릭으로 정량화된 오차를 거의 없이 DP-SGD 및 관련 알고리즘의 전체 프라이버시 프로파일을 포착할 수 있음을 보여줍니다. 우리의 주장을 뒷받침하기 위해, 최신 DP 대규모 이미지 분류와 미국 10년 인구조사를 위한 TopDown 알고리즘의 프라이버시 프로파일을 조사하였으며, 모든 경우에서 GDP가 해당 프로파일에 현저히 잘 맞음을 관찰하였습니다. 마지막으로 이 접근 방식의 강점과 약점에 대해 논의하고, GDP로부터 이점을 얻을 수 있는 다른 프라이버시 메커니즘에 대해 논의합니다.
Webly supervised learning becomes attractive recently for its efficiency in data expansion without expensive human labeling. However, adopting search queries or hashtags as web labels of images for training brings massive noise that degrades the performance of DNNs. Especially, due to the semantic confusion of query words, the images retrieved by one query may contain tremendous images belonging to other concepts. For example, searching `tiger cat' on Flickr will return a dominating number of tiger images rather than the cat images. These realistic noisy samples usually have clear visual semantic clusters in the visual space that mislead DNNs from learning accurate semantic labels. To correct real-world noisy labels, expensive human annotations seem indispensable. Fortunately, we find that metadata can provide extra knowledge to discover clean web labels in a labor-free fashion, making it feasible to automatically provide correct semantic guidance among the massive label-noisy web data. In this paper, we propose an automatic label corrector VSGraph-LC based on the visual-semantic graph. VSGraph-LC starts from anchor selection referring to the semantic similarity between metadata and correct label concepts, and then propagates correct labels from anchors on a visual graph using graph neural network (GNN). Experiments on realistic webly supervised learning datasets Webvision-1000 and NUS-81-Web show the effectiveness and robustness of VSGraph-LC. Moreover, VSGraph-LC reveals its advantage on the open-set validation set.
The COVID-19 pandemic has highlighted the need for a tool to speed up triage in ultrasound scans and provide clinicians with fast access to relevant information. The proposed video-summarization technique is a step in this direction that provides clinicians access to relevant key-frames from a given ultrasound scan (such as lung ultrasound) while reducing resource, storage and bandwidth requirements. We propose a new unsupervised reinforcement learning (RL) framework with novel rewards that facilitates unsupervised learning avoiding tedious and impractical manual labelling for summarizing ultrasound videos to enhance its utility as a triage tool in the emergency department (ED) and for use in telemedicine. Using an attention ensemble of encoders, the high dimensional image is projected into a low dimensional latent space in terms of: a) reduced distance with a normal or abnormal class (classifier encoder), b) following a topology of landmarks (segmentation encoder), and c) the distance or topology agnostic latent representation (convolutional autoencoders). The decoder is implemented using a bi-directional long-short term memory (Bi-LSTM) which utilizes the latent space representation from the encoder. Our new paradigm for video summarization is capable of delivering classification labels and segmentation of key landmarks for each of the summarized keyframes. Validation is performed on lung ultrasound (LUS) dataset, that typically represent potential use cases in telemedicine and ED triage acquired from different medical centers across geographies (India, Spain and Canada).
Quantifying the data uncertainty in learning tasks is often done by learning a prediction interval or prediction set of the label given the input. Two commonly desired properties for learned prediction sets are \emph{valid coverage} and \emph{good efficiency} (such as low length or low cardinality). Conformal prediction is a powerful technique for learning prediction sets with valid coverage, yet by default its conformalization step only learns a single parameter, and does not optimize the efficiency over more expressive function classes. In this paper, we propose a generalization of conformal prediction to multiple learnable parameters, by considering the constrained empirical risk minimization (ERM) problem of finding the most efficient prediction set subject to valid empirical coverage. This meta-algorithm generalizes existing conformal prediction algorithms, and we show that it achieves approximate valid population coverage and near-optimal efficiency within class, whenever the function class in the conformalization step is low-capacity in a certain sense. Next, this ERM problem is challenging to optimize as it involves a non-differentiable coverage constraint. We develop a gradient-based algorithm for it by approximating the original constrained ERM using differentiable surrogate losses and Lagrangians. Experiments show that our algorithm is able to learn valid prediction sets and improve the efficiency significantly over existing approaches in several applications such as prediction intervals with improved length, minimum-volume prediction sets for multi-output regression, and label prediction sets for image classification.
The error-backpropagation (backprop) algorithm remains the most common solution to the credit assignment problem in artificial neural networks. In neuroscience, it is unclear whether the brain could adopt a similar strategy to correctly modify its synapses. Recent models have attempted to bridge this gap while being consistent with a range of experimental observations. However, these models are either unable to effectively backpropagate error signals across multiple layers or require a multi-phase learning process, neither of which are reminiscent of learning in the brain. Here, we introduce a new model, Bursting Cortico-Cortical Networks (BurstCCN), which solves these issues by integrating known properties of cortical networks namely bursting activity, short-term plasticity (STP) and dendrite-targeting interneurons. BurstCCN relies on burst multiplexing via connection-type-specific STP to propagate backprop-like error signals within deep cortical networks. These error signals are encoded at distal dendrites and induce burst-dependent plasticity as a result of excitatory-inhibitory top-down inputs. First, we demonstrate that our model can effectively backpropagate errors through multiple layers using a single-phase learning process. Next, we show both empirically and analytically that learning in our model approximates backprop-derived gradients. Finally, we demonstrate that our model is capable of learning complex image classification tasks (MNIST and CIFAR-10). Overall, our results suggest that cortical features across sub-cellular, cellular, microcircuit and systems levels jointly underlie single-phase efficient deep learning in the brain.
Self-supervised learning of deep neural networks has become a prevalent paradigm for learning representations that transfer to a variety of downstream tasks. Similar to proposed models of the ventral stream of biological vision, it is observed that these networks lead to a separation of category manifolds in the representations of the penultimate layer. Although this observation matches the manifold hypothesis of representation learning, current self-supervised approaches are limited in their ability to explicitly model this manifold. Indeed, current approaches often only apply augmentations from a pre-specified set of "positive pairs" during learning. In this work, we propose a contrastive learning approach that directly models the latent manifold using Lie group operators parameterized by coefficients with a sparsity-promoting prior. A variational distribution over these coefficients provides a generative model of the manifold, with samples which provide feature augmentations applicable both during contrastive training and downstream tasks. Additionally, learned coefficient distributions provide a quantification of which transformations are most likely at each point on the manifold while preserving identity. We demonstrate benefits in self-supervised benchmarks for image datasets, as well as a downstream semi-supervised task. In the former case, we demonstrate that the proposed methods can effectively apply manifold feature augmentations and improve learning both with and without a projection head. In the latter case, we demonstrate that feature augmentations sampled from learned Lie group operators can improve classification performance when using few labels.
Can we develop a model that can synthesize realistic speech directly from a latent space, without explicit conditioning? Despite several efforts over the last decade, previous adversarial and diffusion-based approaches still struggle to achieve this, even on small-vocabulary datasets. To address this, we propose AudioStyleGAN (ASGAN) -- a generative adversarial network for unconditional speech synthesis tailored to learn a disentangled latent space. Building upon the StyleGAN family of image synthesis models, ASGAN maps sampled noise to a disentangled latent vector which is then mapped to a sequence of audio features so that signal aliasing is suppressed at every layer. To successfully train ASGAN, we introduce a number of new techniques, including a modification to adaptive discriminator augmentation which probabilistically skips discriminator updates. We apply it on the small-vocabulary Google Speech Commands digits dataset, where it achieves state-of-the-art results in unconditional speech synthesis. It is also substantially faster than existing top-performing diffusion models. We confirm that ASGAN's latent space is disentangled: we demonstrate how simple linear operations in the space can be used to perform several tasks unseen during training. Specifically, we perform evaluations in voice conversion, speech enhancement, speaker verification, and keyword classification. Our work indicates that GANs are still highly competitive in the unconditional speech synthesis landscape, and that disentangled latent spaces can be used to aid generalization to unseen tasks. Code, models, samples: https://github.com/RF5/simple-asgan/