PyTorchKR Papers

“Image Classification”

총 18,403편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:1406.0312

Generalized Max Pooling

State-of-the-art patch-based image representations involve a pooling operation that aggregates statistics computed from local descriptors. Standard pooling operations include sum- and max-pooling. Sum-pooling lacks discriminability because the resulting representation is strongly influenced by frequent yet often uninformative descriptors, but only weakly influenced by rare yet potentially highly-informative ones. Max-pooling equalizes the influence of frequent and rare descriptors but is only applicable to representations that rely on count statistics, such as the bag-of-visual-words (BOV) and its soft- and sparse-coding extensions. We propose a novel pooling mechanism that achieves the same effect as max-pooling but is applicable beyond the BOV and especially to the state-of-the-art Fisher Vector -- hence the name Generalized Max Pooling (GMP). It involves equalizing the similarity between each patch and the pooled representation, which is shown to be equivalent to re-weighting the per-patch statistics. We show on five public image classification benchmarks that the proposed GMP can lead to significant performance gains with respect to heuristic alternatives.

arXiv:1809.01410

Generating Highly Realistic Images of Skin Lesions with GANs

As many other machine learning driven medical image analysis tasks, skin image analysis suffers from a chronic lack of labeled data and skewed class distributions, which poses problems for the training of robust and well-generalizing models. The ability to synthesize realistic looking images of skin lesions could act as a reliever for the aforementioned problems. Generative Adversarial Networks (GANs) have been successfully used to synthesize realistically looking medical images, however limited to low resolution, whereas machine learning models for challenging tasks such as skin lesion segmentation or classification benefit from much higher resolution data. In this work, we successfully synthesize realistically looking images of skin lesions with GANs at such high resolution. Therefore, we utilize the concept of progressive growing, which we both quantitatively and qualitatively compare to other GAN architectures such as the DCGAN and the LAPGAN. Our results show that with the help of progressive growing, we can synthesize highly realistic dermoscopic images of skin lesions that even expert dermatologists find hard to distinguish from real ones.

arXiv:2007.09859

Novel Approach to Use HU Moments with Image Processing Techniques for Real Time Sign Language Communication

Sign language is the fundamental communication method among people who suffer from speech and hearing defects. The rest of the world doesn't have a clear idea of sign language. "Sign Language Communicator" (SLC) is designed to solve the language barrier between the sign language users and the rest of the world. The main objective of this research is to provide a low cost affordable method of sign language interpretation. This system will also be very useful to the sign language learners as they can practice the sign language. During the research available human computer interaction techniques in posture recognition was tested and evaluated. A series of image processing techniques with Hu-moment classification was identified as the best approach. To improve the accuracy of the system, a new approach height to width ratio filtration was implemented along with Hu-moments. System is able to recognize selected Sign Language signs with the accuracy of 84% without a controlled background with small light adjustments

arXiv:2103.10158공식 코드 · Papers with Code

TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentation

Automatic augmentation methods have recently become a crucial pillar for strong model performance in vision tasks. While existing automatic augmentation methods need to trade off simplicity, cost and performance, we present a most simple baseline, TrivialAugment, that outperforms previous methods for almost free. TrivialAugment is parameter-free and only applies a single augmentation to each image. Thus, TrivialAugment's effectiveness is very unexpected to us and we performed very thorough experiments to study its performance. First, we compare TrivialAugment to previous state-of-the-art methods in a variety of image classification scenarios. Then, we perform multiple ablation studies with different augmentation spaces, augmentation methods and setups to understand the crucial requirements for its performance. Additionally, we provide a simple interface to facilitate the widespread adoption of automatic augmentation methods, as well as our full code base for reproducibility. Since our work reveals a stagnation in many parts of automatic augmentation research, we end with a short proposal of best practices for sustained future progress in automatic augmentation methods.

arXiv:2105.15089

Analogous to Evolutionary Algorithm: Designing a Unified Sequence Model

Inspired by biological evolution, we explain the rationality of Vision Transformer by analogy with the proven practical Evolutionary Algorithm (EA) and derive that both of them have consistent mathematical representation. Analogous to the dynamic local population in EA, we improve the existing transformer structure and propose a more efficient EAT model, and design task-related heads to deal with different tasks more flexibly. Moreover, we introduce the spatial-filling curve into the current vision transformer to sequence image data into a uniform sequential format. Thus we can design a unified EAT framework to address multi-modal tasks, separating the network architecture from the data format adaptation. Our approach achieves state-of-the-art results on the ImageNet classification task compared with recent vision transformer works while having smaller parameters and greater throughput. We further conduct multi-modal tasks to demonstrate the superiority of the unified EAT, e.g., Text-Based Image Retrieval, and our approach improves the rank-1 by +3.7 points over the baseline on the CSS dataset.

arXiv:2201.04309공식 코드 · Papers with Code

Robust Contrastive Learning against Noisy Views

Contrastive learning relies on an assumption that positive pairs contain related views, e.g., patches of an image or co-occurring multimodal signals of a video, that share certain underlying information about an instance. But what if this assumption is violated? The literature suggests that contrastive learning produces suboptimal representations in the presence of noisy views, e.g., false positive pairs with no apparent shared information. In this work, we propose a new contrastive loss function that is robust against noisy views. We provide rigorous theoretical justifications by showing connections to robust symmetric losses for noisy binary classification and by establishing a new contrastive bound for mutual information maximization based on the Wasserstein distance measure. The proposed loss is completely modality-agnostic and a simple drop-in replacement for the InfoNCE loss, which makes it easy to apply to existing contrastive frameworks. We show that our approach provides consistent improvements over the state-of-the-art on image, video, and graph contrastive learning benchmarks that exhibit a variety of real-world noise patterns.

arXiv:2307.06304

Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolution

The ubiquitous and demonstrably suboptimal choice of resizing images to a fixed resolution before processing them with computer vision models has not yet been successfully challenged. However, models such as the Vision Transformer (ViT) offer flexible sequence-based modeling, and hence varying input sequence lengths. We take advantage of this with NaViT (Native Resolution ViT) which uses sequence packing during training to process inputs of arbitrary resolutions and aspect ratios. Alongside flexible model usage, we demonstrate improved training efficiency for large-scale supervised and contrastive image-text pretraining. NaViT can be efficiently transferred to standard tasks such as image and video classification, object detection, and semantic segmentation and leads to improved results on robustness and fairness benchmarks. At inference time, the input resolution flexibility can be used to smoothly navigate the test-time cost-performance trade-off. We believe that NaViT marks a departure from the standard, CNN-designed, input and modelling pipeline used by most computer vision models, and represents a promising direction for ViTs.

arXiv:2501.01406

nnY-Net: Swin-NeXt with Cross-Attention for 3D Medical Images Segmentation

This paper provides a novel 3D medical image segmentation model structure called nnY-Net. This name comes from the fact that our model adds a cross-attention module at the bottom of the U-net structure to form a Y structure. We integrate the advantages of the two latest SOTA models, MedNeXt and SwinUNETR, and use Swin Transformer as the encoder and ConvNeXt as the decoder to innovatively design the Swin-NeXt structure. Our model uses the lowest-level feature map of the encoder as Key and Value and uses patient features such as pathology and treatment information as Query to calculate the attention weights in a Cross Attention module. Moreover, we simplify some pre- and post-processing as well as data enhancement methods in 3D image segmentation based on the dynUnet and nnU-net frameworks. We integrate our proposed Swin-NeXt with Cross-Attention framework into this framework. Last, we construct a DiceFocalCELoss to improve the training efficiency for the uneven data convergence of voxel classification.

arXiv:2210.06583공식 코드 · Papers with Code

S4ND: Modeling Images and Videos as Multidimensional Signals Using State Spaces

Visual data such as images and videos are typically modeled as discretizations of inherently continuous, multidimensional signals. Existing continuous-signal models attempt to exploit this fact by modeling the underlying signals of visual (e.g., image) data directly. However, these models have not yet been able to achieve competitive performance on practical vision tasks such as large-scale image and video classification. Building on a recent line of work on deep state space models (SSMs), we propose S4ND, a new multidimensional SSM layer that extends the continuous-signal modeling ability of SSMs to multidimensional data including images and videos. We show that S4ND can model large-scale visual data in $1$D, $2$D, and $3$D as continuous multidimensional signals and demonstrates strong performance by simply swapping Conv2D and self-attention layers with S4ND layers in existing state-of-the-art models. On ImageNet-1k, S4ND exceeds the performance of a Vision Transformer baseline by $1.5\%$ when training with a $1$D sequence of patches, and matches ConvNeXt when modeling images in $2$D. For videos, S4ND improves on an inflated $3$D ConvNeXt in activity classification on HMDB-51 by $4\%$. S4ND implicitly learns global, continuous convolutional kernels that are resolution invariant by construction, providing an inductive bias that enables generalization across multiple resolutions. By developing a simple bandlimiting modification to S4 to overcome aliasing, S4ND achieves strong zero-shot (unseen at training time) resolution performance, outperforming a baseline Conv2D by $40\%$ on CIFAR-10 when trained on $8 \times 8$ and tested on $32 \times 32$ images. When trained with progressive resizing, S4ND comes within $\sim 1\%$ of a high-resolution model while training $22\%$ faster.

arXiv:2309.06528

Strong-Weak Integrated Semi-supervision for Unsupervised Single and Multi Target Domain Adaptation

Unsupervised domain adaptation (UDA) focuses on transferring knowledge learned in the labeled source domain to the unlabeled target domain. Despite significant progress that has been achieved in single-target domain adaptation for image classification in recent years, the extension from single-target to multi-target domain adaptation is still a largely unexplored problem area. In general, unsupervised domain adaptation faces a major challenge when attempting to learn reliable information from a single unlabeled target domain. Increasing the number of unlabeled target domains further exacerbate the problem rather significantly. In this paper, we propose a novel strong-weak integrated semi-supervision (SWISS) learning strategy for image classification using unsupervised domain adaptation that works well for both single-target and multi-target scenarios. Under the proposed SWISS-UDA framework, a strong representative set with high confidence but low diversity target domain samples and a weak representative set with low confidence but high diversity target domain samples are updated constantly during the training process. Both sets are fused to generate an augmented strong-weak training batch with pseudo-labels to train the network during every iteration. The extension from single-target to multi-target domain adaptation is accomplished by exploring the class-wise distance relationship between domains and replacing the strong representative set with much stronger samples from peer domains via peer scaffolding. Moreover, a novel adversarial logit loss is proposed to reduce the intra-class divergence between source and target domains, which is back-propagated adversarially with a gradient reverse layer between the classifier and the rest of the network. Experimental results based on three benchmarks, Office-31, Office-Home, and DomainNet, show the effectiveness of the proposed SWISS framework.

arXiv:2504.17935

Masked strategies for images with small objects

The hematology analytics used for detection and classification of small blood components is a significant challenge. In particular, when objects exists as small pixel-sized entities in a large context of similar objects. Deep learning approaches using supervised models with pre-trained weights, such as residual networks and vision transformers have demonstrated success for many applications. Unfortunately, when applied to images outside the domain of learned representations, these methods often result with less than acceptable performance. A strategy to overcome this can be achieved by using self-supervised models, where representations are learned and weights are then applied for downstream applications. Recently, masked autoencoders have proven to be effective to obtain representations that captures global context information. By masking regions of an image and having the model learn to reconstruct both the masked and non-masked regions, weights can be used for various applications. However, if the sizes of the objects in images are less than the size of the mask, the global context information is lost, making it almost impossible to reconstruct the image. In this study, we investigated the effect of mask ratios and patch sizes for blood components using a MAE to obtain learned ViT encoder representations. We then applied the encoder weights to train a U-Net Transformer for semantic segmentation to obtain both local and global contextual information. Our experimental results demonstrates that both smaller mask ratios and patch sizes improve the reconstruction of images using a MAE. We also show the results of semantic segmentation with and without pre-trained weights, where smaller-sized blood components benefited with pre-training. Overall, our proposed method offers an efficient and effective strategy for the segmentation and classification of small objects.

arXiv:1911.03451

Enabling Highly Efficient Capsule Networks Processing Through A PIM-Based Architecture Design

In recent years, the CNNs have achieved great successes in the image processing tasks, e.g., image recognition and object detection. Unfortunately, traditional CNN's classification is found to be easily misled by increasingly complex image features due to the usage of pooling operations, hence unable to preserve accurate position and pose information of the objects. To address this challenge, a novel neural network structure called Capsule Network has been proposed, which introduces equivariance through capsules to significantly enhance the learning ability for image segmentation and object detection. Due to its requirement of performing a high volume of matrix operations, CapsNets have been generally accelerated on modern GPU platforms that provide highly optimized software library for common deep learning tasks. However, based on our performance characterization on modern GPUs, CapsNets exhibit low efficiency due to the special program and execution features of their routing procedure, including massive unshareable intermediate variables and intensive synchronizations, which are very difficult to optimize at software level. To address these challenges, we propose a hybrid computing architecture design named \textit{PIM-CapsNet}. It preserves GPU's on-chip computing capability for accelerating CNN types of layers in CapsNet, while pipelining with an off-chip in-memory acceleration solution that effectively tackles routing procedure's inefficiency by leveraging the processing-in-memory capability of today's 3D stacked memory. Using routing procedure's inherent parallellization feature, our design enables hierarchical improvements on CapsNet inference efficiency through minimizing data movement and maximizing parallel processing in memory.

arXiv:2606.18008v2cs.CV

PhaseWin: 충실한 시각적 귀속을 위한 효율적인 검색 알고리즘

PhaseWin: An Efficient Search Algorithm for Faithful Visual Attribution

시각적 귀속(visual attribution)은 현대 비전(vision) 및 비전-언어(vision-language) 모델을 해석하는 기본 도구로, 특히 모델의 결정을 검사, 진단 또는 감사해야 할 때 중요합니다. 그 목표는 모델의 결정이 시각적 입력의 지역적 영역에 어떻게 의존하는지 설명하는 것이며, 일반적으로 후보 이미지 영역에 중요도 순서를 할당함으로써 이루어집니다. $n$개의 영역으로 분할된 이미지가 주어지면, 충실한 귀속은 순서가 있는 부분집합 검색(ordered subset-search) 문제로 공식화할 수 있습니다. 이 문제에서 선택된 영역을 점진적으로 삽입할 때 목표 모델의 응답을 가능한 한 빨리 복원해야 합니다. 영역 부분집합에 대한 전수 탐색(exhaustive search)은 지수적인 비용이 드는 반면, 널리 사용되는 탐욕적 검색(greedy search)은 모든 선택 단계에서 남은 모든 후보를 다시 점수화해야 하므로 여전히 제곱 수준의 모델 평가가 필요합니다. 본 논문에서는 충실한 시각적 귀속을 위한 효율적인 부분집합 검색 알고리즘인 PhaseWin을 제안합니다. PhaseWin은 탐욕적 영역 선택을 단계별 윈도우 검색 절차로 재구성합니다. 매 단계마다 전체 후보 집합을 재평가하는 대신, 전역 후보 스크리닝, 적응형 가지치기(pruning), 지역화된 윈도우 정제(refinement)를 번갈아 수행하면서 탐욕적 검색의 본질적인 영역 순위 매기기 동작을 보존합니다. 우리는 단조 증거 누적(monotone evidence-accumulation) 조건 하에서 PhaseWin을 분석하고, 특징(feature) 수준의 구조적 가정 하에서 제어 가능한 선형 평가 복잡도와 거의 탐욕적(near-greedy)인 충실도 보장을 달성함을 보여줍니다. 이미지 분류, 객체 검출, 시각적 그라운딩(visual grounding), 이미지 캡셔닝(image captioning)에 대한 광범위한 실험을 통해, 비교된 모든 귀속 방법 중에서 PhaseWin이 가장 적은 순방향 패스(forward pass)로 높은 충실도에 도달함을 보여주며, $O(n^2)$에서 $O(n)$으로의 예측된 감소를 경험적으로 실현합니다. 코드는 https://github.com/Qihuai27/phasewin-va 에서 확인할 수 있습니다.

arXiv:2606.18869v1cs.CV

전립선 DWI 보정을 위한 왜곡 학습: 약한 지도 이미지 품질 전이

Learning to Distort: Weakly-Supervised Image Quality Transfer for Prostate DWI Correction

단일 샷 에코 플래너 전립선 확산 강조 이미징(DWI)은 기하학적 왜곡으로 인해 자주 복잡해지며, 이는 해당 이미지에서 신뢰할 수 있는 진단을 도출하는 능력에 영향을 미칩니다. 왜곡된 스캔과 비왜곡 스캔이 쌍으로 존재하는 임상 데이터가 부족하기 때문에 자동화된 보정 방법을 개발하는 것은 어려운 과제입니다. 본 논문에서 우리는 먼저 전이 과정을 지도하기 위해 이미지 품질 평가(IQA) 신호를 활용하는, 비왜곡 이미지에서 왜곡 이미지로의 새로운 약한 지도 이미지 품질 전이(IQT) 프레임워크를 제안합니다. 비용이 많이 드는 복셀 단위 쌍 데이터를 요구하거나 비쌍(unpaired) 알고리즘 개발에 의존하는 기존 방법과 달리, 우리의 접근 방식은 이미지 수준의 품질 라벨(여기서는 왜곡 대 비왜곡)을 활용하여 사전학습된 특징 공간 내에 잠재적 품질 프로토타입을 설정합니다. 현실적인 왜곡을 시뮬레이션하는 것이 직접적인 비쌍 보정보다 더 신뢰할 수 있다는 점을 인식하여, 우리는 생성 궤적을 왜곡된 프로토타입 쪽으로 명시적으로 정규화하여 임상적 열화를 모방하는 현실적인 자기 감도 아티팩트를 생성하는 약한 지도 프로토타입 플로우 매칭 알고리즘을 설명합니다. 이러한 현실적인 쌍을 합성함으로써, 우리는 왜곡 보정을 위해 순방향으로 두 번째 IQT 모델을 학습시킬 수 있게 합니다. 실험 결과, 우리가 생성한 이미지가 실제 아티팩트의 진단 방해를 성공적으로 모방함으로써, 더욱 유능한 왜곡 보정 IQT 모델로 이어진다는 것을 보여줍니다. 정성적 비교 외에도, 우리는 인분포(in-distribution) 및 외부 데이터셋을 사용하여 PI-RADS 및 글리슨 점수 분류에서의 임상 하류 작업 성능을 평가함으로써, 기존 비쌍 접근법(예: CycleGAN, UNIT-DDPM 및 OT-FM)과 우리의 접근법을 순방향 또는 역방향 대안으로서 비교하는 철저한 정량적 평가를 수행합니다.

arXiv:1503.06699

Video-Based Action Recognition Using Rate-Invariant Analysis of Covariance Trajectories

Statistical classification of actions in videos is mostly performed by extracting relevant features, particularly covariance features, from image frames and studying time series associated with temporal evolutions of these features. A natural mathematical representation of activity videos is in form of parameterized trajectories on the covariance manifold, i.e. the set of symmetric, positive-definite matrices (SPDMs). The variable execution-rates of actions implies variable parameterizations of the resulting trajectories, and complicates their classification. Since action classes are invariant to execution rates, one requires rate-invariant metrics for comparing trajectories. A recent paper represented trajectories using their transported square-root vector fields (TSRVFs), defined by parallel translating scaled-velocity vectors of trajectories to a reference tangent space on the manifold. To avoid arbitrariness of selecting the reference and to reduce distortion introduced during this mapping, we develop a purely intrinsic approach where SPDM trajectories are represented by redefining their TSRVFs at the starting points of the trajectories, and analyzed as elements of a vector bundle on the manifold. Using a natural Riemannain metric on vector bundles of SPDMs, we compute geodesic paths and geodesic distances between trajectories in the quotient space of this vector bundle, with respect to the re-parameterization group. This makes the resulting comparison of trajectories invariant to their re-parameterization. We demonstrate this framework on two applications involving video classification: visual speech recognition or lip-reading and hand-gesture recognition. In both cases we achieve results either comparable to or better than the current literature.

arXiv:2103.06232

Quantum machine learning with differential privacy

Quantum machine learning (QML) can complement the growing trend of using learned models for a myriad of classification tasks, from image recognition to natural speech processing. A quantum advantage arises due to the intractability of quantum operations on a classical computer. Many datasets used in machine learning are crowd sourced or contain some private information. To the best of our knowledge, no current QML models are equipped with privacy-preserving features, which raises concerns as it is paramount that models do not expose sensitive information. Thus, privacy-preserving algorithms need to be implemented with QML. One solution is to make the machine learning algorithm differentially private, meaning the effect of a single data point on the training dataset is minimized. Differentially private machine learning models have been investigated, but differential privacy has yet to be studied in the context of QML. In this study, we develop a hybrid quantum-classical model that is trained to preserve privacy using differentially private optimization algorithm. This marks the first proof-of-principle demonstration of privacy-preserving QML. The experiments demonstrate that differentially private QML can protect user-sensitive information without diminishing model accuracy. Although the quantum model is simulated and tested on a classical computer, it demonstrates potential to be efficiently implemented on near-term quantum devices (noisy intermediate-scale quantum [NISQ]). The approach's success is illustrated via the classification of spatially classed two-dimensional datasets and a binary MNIST classification. This implementation of privacy-preserving QML will ensure confidentiality and accurate learning on NISQ technology.

arXiv:2506.04673

Interpretable Few-Shot Image Classification via Prototypical Concept-Guided Mixture of LoRA Experts

Self-Explainable Models (SEMs) rely on Prototypical Concept Learning (PCL) to enable their visual recognition processes more interpretable, but they often struggle in data-scarce settings where insufficient training samples lead to suboptimal performance.To address this limitation, we propose a Few-Shot Prototypical Concept Classification (FSPCC) framework that systematically mitigates two key challenges under low-data regimes: parametric imbalance and representation misalignment. Specifically, our approach leverages a Mixture of LoRA Experts (MoLE) for parameter-efficient adaptation, ensuring a balanced allocation of trainable parameters between the backbone and the PCL module.Meanwhile, cross-module concept guidance enforces tight alignment between the backbone's feature representations and the prototypical concept activation patterns.In addition, we incorporate a multi-level feature preservation strategy that fuses spatial and semantic cues across various layers, thereby enriching the learned representations and mitigating the challenges posed by limited data availability.Finally, to enhance interpretability and minimize concept overlap, we introduce a geometry-aware concept discrimination loss that enforces orthogonality among concepts, encouraging more disentangled and transparent decision boundaries.Experimental results on six popular benchmarks (CUB-200-2011, mini-ImageNet, CIFAR-FS, Stanford Cars, FGVC-Aircraft, and DTD) demonstrate that our approach consistently outperforms existing SEMs by a notable margin, with 4.2%-8.7% relative gains in 5-way 5-shot classification.These findings highlight the efficacy of coupling concept learning with few-shot adaptation to achieve both higher accuracy and clearer model interpretability, paving the way for more transparent visual recognition systems.

arXiv:2606.22400v1cs.CV

전이학습을 활용한 계산 효율적인 CNN 기반 다암종 검출

Multi-cancer detection using a computationally efficient CNN with transfer learning

본 연구는 생의학 영상을 이용한 다암종 검출을 위해 전이학습으로 강화된 계산 효율적인 합성곱 신경망(CNN) 아키텍처를 제안한다. 제안된 경량 CNN 모델은 높은 분류 성능을 유지하면서 계산 복잡도를 줄이도록 설계되어, 자원이 제한된 환경에서의 배포에 적합하다. 본 접근법은 뇌 자기공명영상(MRI)과 폐·신장 컴퓨터 단층촬영(CT) 스캔으로 구성된 세 가지 서로 다른 종양 데이터셋에서 평가하였다. 5-fold 계층화 교차검증(CV)을 사용한 결과, 뇌·폐·신장 암 분류에 대해 각각 90.85 ± 2.22%, 98.64 ± 2.43%, 99.92 ± 0.08%의 테스트 정확도를 달성하였다. 전이학습은 한 암종 유형으로 모델을 사전학습한 뒤 다른 암종에 미세조정하는 방식으로 적용되며, 처음부터 학습한 모델과 비슷한 성능을 달성하는 데 추가 20 에포크만 필요하다. 미세조정 과정에서는 CNN의 분류 부분을 갱신하며, NVIDIA GeForce GTX 960에서 에포크당 이미지당 약 0.014초가 소요된다. 비교 평가 결과, 제안 모델은 Xception, VGG16, VGG19, MobileNetV2, DenseNet121 등 여러 최신 사전학습 아키텍처보다 우수한 성능을 보였다. 전반적으로 본 모델의 효과는 형태학적 특성이 서로 다른 세 가지 암종에 걸쳐 평가되었으며, MRI와 CT 영상 방식 모두에서의 성능을 검증하고 다양한 과제와 데이터 유형에 걸친 견고한 성능을 입증하였다. 이러한 결과는 계산 자원이 제한된 환경에서 특히 정확도를 희생하지 않고 암 진단을 가속화하는 데 있어 간결한 심층 학습(DL) 프레임워크의 잠재력을 강조한다.

arXiv:1808.06675

Class2Str: End to End Latent Hierarchy Learning

Deep neural networks for image classification typically consists of a convolutional feature extractor followed by a fully connected classifier network. The predicted and the ground truth labels are represented as one hot vectors. Such a representation assumes that all classes are equally dissimilar. However, classes have visual similarities and often form a hierarchy. Learning this latent hierarchy explicitly in the architecture could provide invaluable insights. We propose an alternate architecture to the classifier network called the Latent Hierarchy (LH) Classifier and an end to end learned Class2Str mapping which discovers a latent hierarchy of the classes. We show that for some of the best performing architectures on CIFAR and Imagenet datasets, the proposed replacement and training by LH classifier recovers the accuracy, with a fraction of the number of parameters in the classifier part. Compared to the previous work of HDCNN, which also learns a 2 level hierarchy, we are able to learn a hierarchy at an arbitrary number of levels as well as obtain an accuracy improvement on the Imagenet classification task over them. We also verify that many visually similar classes are grouped together, under the learnt hierarchy.

arXiv:1811.06488

Exploring the Deep Feature Space of a Cell Classification Neural Network

In this paper, we present contemporary techniques for visualising the feature space of a deep learning image classification neural network. These techniques are viewed in the context of a feed-forward network trained to classify low resolution fluorescence images of white blood cells captured using optofluidic imaging. The model has two output classes corresponding to two different cell types, which are often difficult to distinguish by eye. This paper has two major sections. The first looks to develop the information space presented by dimension reduction techniques, such as t-SNE, used to embed high-dimensional pre-softmax layer activations into a two-dimensional plane. The second section looks at feature visualisation by optimisation to generate feature images representing the learned features of the network. Using and developing these techniques we visualise class separation and structures within the dataset at various depths using clustering algorithms and feature images; track the development of feature complexity as we ascend the network; and begin to extract the features the network has learnt by modulating single-channel feature images with up-scaled neuron activation maps to distinguish their most salient parts.