Intel Updates and Fixes Vulnerability in NeuralCompressor

TapTechNews on May 21, Intel's monthly routine update fixed a total of 41 vulnerabilities. Different from only fixing processor vulnerabilities in the past, this Intel update focused on fixing an elevation of privilege vulnerability CVE-2024-22476 in its open-source Python library NeuralCompressor for model compression.

According to Intel's security bulletin, the vulnerability numbered CVE-2024-22476 has a CVSS rating of full 10 points, mainly related to 'improper input validation'. Hackers can remotely exploit the relevant vulnerability to elevate privileges. Intel has currently launched the NeuralCompressor 2.5.0 version to fix the relevant vulnerability.

TapTechNews learned from querying relevant reports that NeuralCompressor is mainly applied to deep learning deployment on CPUs or GPUs and supports popular model compression techniques on all mainstream deep learning frameworks (TensorFlow, PyTorch, ONNXRuntime, and MXNet), such as quantization, pruning (sparsity), distillation, and neural architecture search. Interested friends can visit the corresponding library page: click here to visit.