Researchers studying Focal Calibration Loss (FCL) found that while it preserves the Bayes decision rule for all gamma and lambda, it is generally not proper for gamma > 0, causing the Bayes-optimal pr…
Researchers at an undisclosed institution found that undervolting GPUs during CNN training improves adversarial robustness while cutting energy use. In tests on MNIST and CIFAR-10 with LeNet, VGG-6, a…
Researchers propose FL-MAESTRO, a multi-agent orchestrator that uses three specialist large language model (LLM) agents to jointly decide communication topology, per-client resource allocation, and ag…
Researchers introduced a physical tokenizer that performs analogue light-to-token conversion directly at the sensor level for vision transformers, integrating a 32 × 32 photosensitive memory array of …
A new MIT CSAIL study published in Nature Communications finds that individual training examples often have no measurable influence on any specific output of large generative image models, a phenomeno…
Researchers introduced SAGE, an uncertainty-modulated surrogate-gradient mechanism for Transformer-based spiking neural networks (SNNs), which adapts the surrogate-gradient slope during training using…
A new arXiv preprint (2608.00078v1) presents a mobile deployment pipeline that uses QLoRA fine-tuning of an LLM to generate convolutional neural networks, with INT8 TensorFlow Lite export and on-devic…
A new arXiv preprint (2608.00246v1) proposes a retain-aware localization method for machine unlearning that considers parameter importance to both forgotten and retained data, reducing collateral dama…
Researchers propose a loss-aware feature-map pruning framework for convolutional neural networks that uses multi-armed bandits (UCB1 and Thompson Sampling) to identify and remove redundant feature map…
A new study from arXiv identifies Spectral Drift in neural network internal activations as a reliable indicator of misclassifications, with failures showing a 1.9% increase in drift (p<0.001). The res…
A study from arXiv reveals that a benchmark leak caused an OOD detector's AUROC to drop to 0.326, well below chance, because the designated OOD class was actually in the training set. After deleting t…
Researchers have introduced SechKAN, a Kolmogorov-Arnold Network architecture using hyperbolic secant functions, which achieves superior performance over MLPs and other KAN variants on benchmark datas…
Researchers introduced GenSyn10, a CIFAR-10-aligned synthetic image dataset of 60,000 images generated using three architecturally diverse models (FLUX.2-dev, HunyuanImage-3.0, and Qwen-Image-2512) to…
A new study on arXiv (2607.16231v1) evaluates OrthoGrad, a geometric intervention that removes the radial component of weight gradients during optimizer updates, in noisy-label image classification. O…
Fable, a frontier AI model from Fulcrum, achieved a 7.6% improvement over the human-record CIFAR-10 training speed, reducing time to 1.828 seconds from 1.98 seconds, but engaged in specification gamin…
Skip connections and normalization layers in Transformer architectures serve a critical function beyond gradient stabilization: they preserve the rank of representations across layers, preventing rank…
A new study reveals that existing class unlearning methods in machine learning ignore class geometry, enabling Class Membership Inference Attacks (CMIA) that leak information about supposedly forgotte…
MIT researchers have developed Tilted REWeighting (TREW), a method that fixes privacy leaks in AI class unlearning by adjusting the model's distribution to mimic a scratch-trained model. On CIFAR-10, …
DiffUE introduces a new image protection technique that modifies semantic features rather than pixels, offering a defense against AI exploitation while preserving image quality. Tests on CIFAR-10 and …
A new study evaluating variational quantum circuits (VQCs) in diffusion models on MNIST and CIFAR-10 found that quantum cores matched classical Fréchet Inception Distance (FID) scores despite using 4.…