AI Papers - 2026-04-15

Today's papers: - NTIRE 2026 The 3rd Restore Any Image Model (RAIM) Challenge: Professional Image Quality Assessment (Track 1): https://arxiv.org/abs/2604.12512v1 - RePAIR: Interactive Machine Unlearning through Prompt-Aware Model Repair: https://arxiv.org/abs/2604.12820v1 - Decoding by Perturbation: Mitigating MLLM Hallucinations via Dynamic Textual Perturbation: https://arxiv.org/abs/2604.12424v1 - Fully Homomorphic Encryption on Llama 3 model for privacy preserving LLM inference: https://arxiv.org/abs/2604.12168v1 - MISID: A Multimodal Multi-turn Dataset for Complex Intent Recognition in Strategic Deception Games: https://arxiv.org/abs/2604.12700v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.