Finding Tiny Faces (by Hu and Ramanan) proposes a novel approach to find small objects in an image. Our contribution consists in deeply understanding the choices of the paper together with applying and extending a similar method to a real world subject which is the counting of people in a public demonstration.
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
168,932 papers · 148 categories
Trend · papers per month
48 results for “tiny images”
TrackNet: A Deep Learning Network for Tracking High-speed and Tiny Objects in Sports Applicationscs.LG
TrackNet tracks tiny balls in sports videos with high precision.
problem Accurately tracking tiny and fast-moving objects like tennis balls in sports videos.
method Developed a heatmap-based deep learning network, TrackNet, trained on public and additional labeled videos.
result Achieved high precision (99.7%) in ball tracking on public domain videos.
Image classifiers are sensitive to small changes, affecting most images in a class.
problem Sensitivity of image classifiers to small perturbations.
method Demonstrated sensitivity for any classifier over images, showing that for most classes, a tiny perturbation can change the classification of a majority of images.
result Image classifiers are sensitive to small perturbations, affecting most images in a class.
Unsupervised training improves adversarial robustness.
problem Training robust models requires large labeled datasets, which are expensive.
method Used unlabeled data to train adversarially robust models.
result Unsupervised Adversarial Training improves robust accuracy by 21.7%.
Tiny Eats GRU detects eating episodes on a microcontroller.
problem Automatic dietary monitoring on low-power devices.
method Shallow gated recurrent unit (GRU) architecture on Arm Cortex M0+.
result Tiny Eats GRU achieves 95.15% accuracy with 4% memory usage and 6 ms latency.
Deep neural networks are vulnerable to adversarial attacks. The literature is rich with algorithms that can easily craft successful adversarial examples. In contrast, the performance of defense techniques still lags behind. This paper proposes ME-Net, a defense method that leverages matrix estimation (ME). In ME-Net, i…
Faster and accurate JPEG2000 image classification without reconstruction.
problem Efficiently classify j2k-compressed images without reconstructing them.
method Train a deep CNN using DWT coefficients directly from j2k-compressed images, using different augmentation techniques.
result Achieved faster and more accurate classification of j2k images without additional computation.
New framework explains adversarial examples in neural nets.
problem Adversarial examples in deep neural networks are counterintuitive and hard to explain.
method Introduces Dimpled Manifold Model to explain training phases and adversarial examples.
result Shows how adversarial examples arise from training dynamics and dimpling phase.
Tiny non-Leighton complexesmath.GT
Tiny complexes share 3-5 triangles in common coverings.
problem Complexes sharing common coverings with finite triangles.
method Pseudo-simplicial triangulation analysis.
result Minimum triangles in common coverings are 3, 4, or 5.
Tab-TRM uses recursive model for insurance pricing on tabular data.
problem Insurance pricing on tabular data.
method Adapts recursive latent reasoning to insurance modeling using a compact, parameter-efficient network.
result Improves insurance pricing accuracy using iterative refinement of latent tokens.
A new clustering algorithm considers data smoothness for better performance.
problem Clustering multi-scale data with varying cluster densities.
method Divide objects into tiny clusters, cluster centers form smooth graphs.
result Significantly outperforms state-of-the-art clustering algorithms.
One Man's Trash is Another Man's Treasure: Resisting Adversarial Examples by Adversarial Examplescs.LG
Transform images with adversarial examples for robust classification.
problem Adversarial examples mislead deep neural networks in image classification.
method Transform input images with adversarial examples from a pre-trained model.
result Significantly more robust against a wide range of attacks compared to state-of-the-art methods.
Study uses deep learning to detect BCCs in high-res histopathological images.
problem Detecting BCCs in high-resolution, weakly labeled histopathological images.
method Attention-based deep learning models to process ultra-high resolution images with weak labels.
result Attention-based models achieve almost perfect classification performance (AUC of 0.99).
This paper represents a preliminary (pre-reviewing) version of a sublinear variational algorithm for isotropic Gaussian mixture models (GMMs). Further developments of the algorithm for GMMs with diagonal covariance matrices (instead of isotropic clusters) and their corresponding benchmarking results have been published…
Over the past decade, deep neural networks (DNNs) have demonstrated remarkable performance in a variety of applications. As we try to solve more advanced problems, increasing demands for computing and power resources has become inevitable. Spiking neural networks (SNNs) have attracted widespread interest as the third-g…
Tiny benchmarks reduce LLM evaluation costs by using fewer examples.
problem Expensive evaluation of LLMs with tens of thousands of examples.
method Developed evaluation tools and tiny versions of popular benchmarks.
result Accurately estimate LLM performance with just 100 curated examples.
Hash codes are efficient data representations for coping with the ever growing amounts of data. In this paper, we introduce a random forest semantic hashing scheme that embeds tiny convolutional neural networks (CNN) into shallow random forests, with near-optimal information-theoretic code aggregation among trees. We s…
New algorithm PRACTISE accelerates networks with tiny sets, reducing latency by 22%.
problem Accelerating networks with tiny training sets to improve performance and privacy.
method Proposes a new algorithm PRACTISE based on finetune convexity hypothesis and feature mimicking.
result PRACTISE outperforms previous methods by reducing latency by 22% on ImageNet-1k.
ThriftyNet uses a single convolutional layer recursively to maximize parameter usage.
problem Maximizing the use of parameters in deep convolutional neural networks.
method A single convolutional layer is used recursively, with normalization, non-linearities, downsampling, and shortcuts to maintain model expressivity.
result ThriftyNet achieves competitive performance with significantly fewer parameters.
Study fusion methods for financial image views to improve robustness against attacks.
problem Improving robustness of financial image views for next-day direction prediction.
method Same-source multi-view learning with early fusion and late fusion, using OHLCV and technical-indicator views, and evaluating pixel-space L-infinity attacks.
result Early fusion can suffer negative transfer under noisy settings, while late fusion is more reliable once labels stabilize.
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.
Unlabeled data improves adversarial robustness in machine learning.
problem Improving robustness of machine learning models against adversarial attacks.
method Semisupervised learning with self-training using unlabeled data.
result Semisupervised learning with unlabeled data significantly improves adversarial robustness.
New algorithm improves dataset distillation with 108% improvement on ImageNet.
problem Improving dataset distillation performance.
method Reparameterization and convexification of implicit gradients (RCIG).
result Establishes new state-of-the-art on various dataset distillation tasks.
Understanding how neural networks learn remains one of the central challenges in machine learning research. From random at the start of training, the weights of a neural network evolve in such a way as to be able to perform a variety of tasks, like classifying images. Here we study the emergence of structure in the wei…
On the variety of Lie algebras endowed with complex structures: degenerations and deformationsmath.RT
Study Lie algebras with complex structures, focusing on degenerations and deformations.
problem Understanding the space of Lie algebras with complex structures and their transformations.
method Identifying invariants that remain consistent under degenerations and applying to four-dimensional case.
result Found invariants that help in understanding the behavior of Lie algebras under complex structures.
Bayes-optimal classifiers are robust to adversarial attacks, unlike CNNs trained on the same data.
problem The vulnerability of modern CNN classifiers to adversarial examples.
method Constructing realistic image datasets and deriving analytic conditions for Bayes-optimal classifiers.
result Bayes-optimal classifiers are robust to adversarial attacks, unlike CNNs trained on the same data.
EAST compresses deep ConvNets for tiny memory nodes.
problem Memory constraints in tiny devices for deep ConvNets.
method Encoding-Aware Sparse Training (EAST) with adaptive group pruning and LZ4 weight encoding.
result EAST achieves deep memory compression with lower sparsity and higher accuracy.
xYOLO speeds up object detection on low-end hardware for humanoid soccer robots.
problem Real-time object detection on resource-constrained devices like Raspberry Pi.
method Adaptation of YOLO CNN model to achieve faster inference speed.
result Achieved 9.66 FPS on Raspberry Pi 3 B, 70x faster than Tiny-YOLO.
Flat space for manifolds with tiny curvature.
problem Understanding manifolds with curvature concentration.
method Analyzing non-compact manifolds with non-negative Ricci curvature and small curvature concentration.
result Manifolds with curvature concentration are flat.
Spider GAN accelerates GAN training with a new approach.
problem Stable training of Generative adversarial networks (GANs).
method Spider GAN leverages a novel approach to identify closely related datasets (friendly neighborhoods) and uses a new measure (signed inception distance) to accelerate GAN training.
result Spider GAN achieves faster convergence and state-of-the-art FID values with one-fifth of the training iterations.
Energy-efficient detection of natural errors in deep networks.
problem Deep networks lack error detection capability without additional energy costs.
method Append RACs at hidden layers to detect natural errors with early classification termination.
result Early classification termination reduces energy consumption.
Improved deep learning model deployment on tiny MCUs with mixed-precision quantization.
problem Memory limitations prevent accurate deployment of DNN models on tiny MCUs.
method Automated mixed-precision quantization using Reinforcement Learning for MCU constraints.
result Mixed-precision models achieve high accuracy with uniform quantization policies.
VarDeepPCA refines medical image segmentation from small datasets, improving anatomical plausibility and reducing errors.
problem Medical image segmentation fails on out-of-distribution data due to variations in scanners and protocols.
method VarDeepPCA learns valid anatomical geometries using only small in-distribution datasets, providing uncertainty estimates.
result VarDeepPCA restores segmentation maps to OOD data, improving anatomical plausibility and reducing errors.
Supervised Dimensionality Reduction Revisited: Why LDA on Frozen CNN Features Deserves a Second Lookcs.LG
LDA improves image classification accuracy with fewer features.
problem Fine-grained image classification with pretrained features.
method Supervised dimensionality reduction with LDA before linear probing.
result LDA improves accuracy over full features in 11 out of 12 configurations.
Puzzle Mix optimizes mixup by leveraging saliency and local statistics.
problem Overfitting and adversarial vulnerability of deep networks.
method Puzzle Mix alternates between multi-label and saliency-based objectives for optimal mixing masks.
result Puzzle Mix outperforms other mixup methods in generalization and adversarial robustness.
New methods improve anomaly detection in deep networks by leveraging hierarchical likelihoods and multi-scale features.
problem Challenges in detecting anomalies in high-level features due to model bias and domain prior.
method Two methods: 1) Log likelihood ratios between in-distribution and general distribution models, 2) Multi-scale likelihood contribution.
result Strong anomaly detection performance in unsupervised setting, slightly underperforming supervised methods.
The combination of deep neural network models and reinforcement learning algorithms can make it possible to learn policies for robotic behaviors that directly read in raw sensory inputs, such as camera images, effectively subsuming both estimation and control into one model. However, real-world applications of reinforc…
Light pillars over rippled water appear parallel due to projection geometry.
problem Light pillars over rippled water appear parallel.
method Developing a geometric optics model
result The phenomenon is explained using the specular reflection rule, projection geometry, and physics of surface slopes.
Develops a method for manifold learning with small sample size datasets.
problem Improving manifold learning performance for multiple tasks with limited samples.
method Uses instance and model transfer to integrate manifold models from similar tasks.
result Successfully estimates manifolds with tiny sample sizes across multiple tasks.
This work analyzes how frequency components affect CNN predictions and robustness.
problem Lack of frequency-based explanation for CNNs leading to vulnerabilities.
method Frequency component analysis and quantification of their contribution to CNN predictions.
result Adversarial attacks exploit high-frequency features, while robustness comes from low-frequency associations.
Softmax is found ineffective for NL block, leading to improved performance.
problem Inefficiency of softmax in NL block for global context modeling.
method Empirical analysis and replacement of softmax with scaling factor.
result Improved performance on various datasets with reduced computational cost.
Proposes deep collective learning to learn inputs and weights together in neural networks.
problem The need to optimize inputs alongside weights in deep neural networks.
method Introduces deep collective learning to learn inputs and weights jointly, using lookup tables.
result Demonstrates advantages and promise of Lookup-VNets and deep collective learning.
ALCORE tensor decomposition reduces computational cost for sparse count data.
problem Efficiently decompose sparse count data matrices.
method Probabilistic Tucker decomposition with -norm constraint.
result ALCORE achieves similar results to full Tucker decomposition at a fraction of the cost.
POET enables large neural network training on tiny devices with reduced energy.
problem Training large neural networks on memory-limited edge devices.
method Jointly optimizes rematerialization and paging for memory reduction, formulating an MILP for energy-efficient training.
result POET trains ResNet-18 and BERT within Cortex-M memory constraints, outperforming current methods in energy efficiency.
Generative model improves scene graph prediction for rare compositions.
problem Challenges in inferring unseen or rare scene graph compositions.
method Synthesizing rare scene graphs by perturbing real ones, using conditional GANs.
result Marginal but consistent improvements in zero- and few-shot metrics.
Improved bound for optimal Moebius band aspect ratio.
problem Finding the optimal aspect ratio of a Moebius band.
method Optimization problem to find the smallest possible aspect ratio of a smoothly embedded Moebius band.
result The new bound is at least , significantly improving on previous bounds.
In continual learning (CL), an agent learns from a stream of tasks leveraging prior experience to transfer knowledge to future tasks. It is an ideal framework to decrease the amount of supervision in the existing learning algorithms. But for a successful knowledge transfer, the learner needs to remember how to perform …
CutMix training technique improves spatial locality in Vision Transformers.
problem Improving spatial locality in Vision Transformers trained from scratch.
method Comparison of Baseline and Modern training protocols on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
result CutMix training component significantly reduces Mean Attention Distance (MAD) in early layers of Vision Transformers.