discrete cosine transform
PulseAugur coverage of discrete cosine transform — every cluster mentioning discrete cosine transform across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
SpecSlot enhances generative Transformers for detailed images with less computation
Researchers have introduced SpecSlot, a novel tokenization method for pixel-space generative Transformers designed to enhance image detail without increasing computational costs. This approach decouples output detail fr…
-
New defense blocks AI prompt injection via image steganography
Researchers have developed a new defense mechanism against indirect prompt injection attacks targeting agentic AI systems that use the Model Context Protocol (MCP). These attacks exploit untrusted tool outputs, such as …
-
New DRiC method improves online residual correction for forecasters
Researchers have developed DRiC (DCT Ridge Corrector), a novel method for online residual correction of frozen multivariate forecasters, particularly effective under label delay conditions. DRiC aims to minimize retaine…
-
New MaRK framework dynamically conditions State Space Models
Researchers have introduced MaRK (Markov-adapted Recurrent Kernels), a novel framework for dynamically conditioning State Space Models (SSMs) in iterative generation tasks. Unlike previous methods that modulated inputs …
-
New EPOC method improves time series forecasting with compressed state
Researchers have developed a new method called Endpoint-Preserving Online Correction (EPOC) for multi-horizon time series forecasting. EPOC addresses the challenge of retaining residual feedback without excessive state …
-
New phase unwrapping method improves image processing accuracy
Researchers have developed a novel translation-invariant tile-based phase unwrapping method that combines frequency-domain unwrapping with spatial merging. This approach jointly performs unwrapping and noise filtering, …
-
TRACE method corrects ring artifacts in photon-counting CT scans
A new method called TRACE has been developed to correct ring artifacts in photon-counting detector computed tomography (PCD-CT) scans. This unsupervised, two-stage sinogram decomposition technique estimates and corrects…
-
New theory offers lower bound for quantized matrix multiplication error
Researchers have developed a new theoretical framework to minimize error in quantized matrix multiplication. The study, published on arXiv, introduces a nuclear-norm lower bound for dithered scalar quantization, providi…
-
New language model advances generative image compression
Researchers have developed FLM, a novel frequency-aware language model designed for generative image compression. This model transforms images into discrete sequences of DCT coefficients, which FLM then uses to autoregr…
-
New lightweight Med-VQA framework uses memory and graph attention
Researchers have developed MedFG-VQA, a novel lightweight framework designed to improve medical Visual Question Answering (Med-VQA) capabilities. This system utilizes a memory bank to enhance low-frequency features deri…
-
New methods estimate image similarity from compression error
Researchers have developed new methods to estimate the Structural Similarity Index (SSIM) from Mean Square Error (MSE) for DCT-based compressed images. These approaches approximate local MSE by redistributing global MSE…
-
New trainable bases improve image compression beyond JPEG
Researchers have developed a new method for image compression using trainable multilinear bases, inspired by quantum many-body theory. These bases, parameterized as isometric tensor networks and optimized on unitary mat…
-
New DDMM Framework Enhances Hyperspectral Image Fusion
Researchers have introduced a novel framework called Dual-Domain Manifold Modeling (DDMM) to improve hyperspectral image fusion. This approach addresses limitations in existing methods by better modeling geometric const…
-
New technique improves Transformer KV cache compression
Researchers have developed Codec-Gauge, a post-training layer designed to improve the compression of Key-Value (KV) caches in long-context Transformer models. This method learns orthogonal channel transforms that optimi…
-
Quantum Kitchen Sinks enhance RF spectrogram anomaly detection · 2 sources tracked
Researchers have developed a novel approach for detecting anomalies in radio-frequency (RF) spectrograms using Quantum Kitchen Sinks (QKS). This method extends the standard QKS template with multi-depth data re-uploadin…
-
Freqformer Transformer tackles image demoiréing with frequency decomposition
Researchers have introduced Freqformer, a novel Transformer-based framework designed to address the challenging task of image demoiréing. This method effectively decomposes moiré patterns into distinct high-frequency te…
-
New lightweight framework enhances underwater images using frequency and spatial data
Researchers have developed a novel, lightweight framework for real-time underwater image enhancement (UIE) that integrates frequency domain information with spatial domain processing. The proposed system utilizes a Mult…
-
FrequencyFormer pipeline boosts vision transformer efficiency for edge devices
Researchers have developed FrequencyFormer, a novel pipeline designed to make vision transformers (ViTs) more efficient for deployment on sensor-edge systems. This approach leverages the frequency domain to compress ima…
-
New FAFM method generates continuous, stable robotic actions
Researchers have developed Frequency-Aware Flow Matching (FAFM), a novel technique to improve robotic action generation by producing continuous and temporally consistent movements. FAFM addresses limitations in existing…
-
New ShearFuse-UNet model predicts wildfire spread efficiently
Researchers have developed ShearFuse-UNet, a novel deep learning model designed for predicting wildfire spread using satellite data. This model is notable for its lightweight architecture and computational efficiency, i…