PulseAugur
LIVE 12:27:45
research · [7 sources] ·
0
research

Researchers explore spatiotemporal convolutions, explainable AI, and backdoor mitigation in new papers

By PulseAugur Editorial · Summary by gemini-2.5-flash-lite from 7 sources

Researchers have explored spatiotemporal convolutions for EEG signal classification, finding that 2D convolutions can significantly reduce training time in high-dimensional tasks while maintaining performance. Separately, a study adapted an explanation technique for Transformer-based genome language models (gLMs) like DNABERT-2, demonstrating that these models can provide biological insights comparable to CNNs. AI

Summary written by gemini-2.5-flash-lite from 7 sources. How we write summaries →

IMPACT Advances in explainability for genome language models and efficient EEG classification could accelerate research in bioinformatics and neuroscience.

RANK_REASON The cluster contains two academic papers discussing novel applications and explainability of neural network architectures.

Read on arXiv cs.LG →

Researchers explore spatiotemporal convolutions, explainable AI, and backdoor mitigation in new papers

COVERAGE [7]

  1. arXiv cs.LG TIER_1 · Laurits Dixen, Stefan Heinrich, Paolo Burelli ·

    Spatiotemporal Convolutions on EEG signal -- A Representation Learning Perspective on Efficient and Explainable EEG Classification with Convolutional Neural Nets

    arXiv:2605.03874v1 Announce Type: new Abstract: Classification of EEG signals using shallow Convolutional Neural Networks (CNNs) is a prevalent and successful approach across a variety of fields. Most of these models use independent one-dimensional (1D) convolutional layers along…

  2. arXiv cs.AI TIER_1 · Paolo Burelli ·

    Spatiotemporal Convolutions on EEG signal -- A Representation Learning Perspective on Efficient and Explainable EEG Classification with Convolutional Neural Nets

    Classification of EEG signals using shallow Convolutional Neural Networks (CNNs) is a prevalent and successful approach across a variety of fields. Most of these models use independent one-dimensional (1D) convolutional layers along the spatial and temporal dimensions, which are …

  3. Hugging Face Daily Papers TIER_1 ·

    Evaluating Post-hoc Explanations of the Transformer-based Genome Language Model DNABERT-2

    Explaining deep neural network predictions on genome sequences enables biological insight and hypothesis generation-often of greater interest than predictive performance alone. While explanations of convolutional neural networks (CNNs) have been shown to capture relevant patterns…

  4. arXiv cs.LG TIER_1 · Bernhard Y. Renard ·

    Evaluating Post-hoc Explanations of the Transformer-based Genome Language Model DNABERT-2

    Explaining deep neural network predictions on genome sequences enables biological insight and hypothesis generation-often of greater interest than predictive performance alone. While explanations of convolutional neural networks (CNNs) have been shown to capture relevant patterns…

  5. arXiv cs.CV TIER_1 · Kealan Dunnett, Reza Arablouei, Dimity Miller, Volkan Dedeoglu, Raja Jurdak ·

    Backdoor Mitigation in Object Detection via Adversarial Fine-Tuning

    arXiv:2605.05928v1 Announce Type: new Abstract: Backdoor attacks can implant malicious behaviours into deep models while preserving performance on clean data, posing a serious threat to safety-critical vision systems. Although backdoor mitigation has been studied extensively for …

  6. arXiv cs.CV TIER_1 · Anjith George, Sebastien Marcel ·

    Lightweight Cross-Spectral Face Recognition via Contrastive Alignment and Distillation

    arXiv:2605.04769v1 Announce Type: new Abstract: Heterogeneous Face Recognition (HFR) aims at matching face images captured across different sensing modalities, such as thermal-to-visible or near-infrared-to-visible, enhancing the usability of face recognition systems in challengi…

  7. arXiv cs.CV TIER_1 · Sebastien Marcel ·

    Lightweight Cross-Spectral Face Recognition via Contrastive Alignment and Distillation

    Heterogeneous Face Recognition (HFR) aims at matching face images captured across different sensing modalities, such as thermal-to-visible or near-infrared-to-visible, enhancing the usability of face recognition systems in challenging real-world conditions. Although recent HFR me…

RELATED ENTITIES

RELATED TOPICS