Loss-Aware Feature-Map Pruning in Convolutional Neural Networks Using Multi-Armed Bandits

AI 中文解读

1. 核心亮点：科学家给AI模型装上“老虎机”算法，能在几乎不损失准确率的情况下，自动剪掉大量冗余计算模块，让AI跑得更快更省资源。 2. 通俗解读：想象一下你家的房间堆满了家具，有些其实根本用不上。卷积神经网络里也有类似问题——很多“通道”（类似信息管道）是多余的，白白占用内存和算力。过去剪掉这些多余通道像蒙眼拆家具，可能不小心拆掉承重墙。这篇论文的方法就像让AI自己“赌博”：它把每个通道当成一台老虎机，先试玩一小局（临时遮掉该通道看模型表现变差多少），如果变差很少就记录为“安全移除奖励”。经过固定次数试玩后，排名靠后的通道连同背后的权重、偏置一并永久删除。整个过程仅用少量数据评估，不需要跑完整测试集。 3. 实际影响：这项技术最直接的好处是让手机、智能音箱等设备上的AI应用更流畅、更省电。比如拍照时的人脸识别、语音助手的实时响应，甚至自动驾驶中的图像处理，都能在保持原有准确率的前提下大幅减少计算量。对普通用户来说，设备不用再频繁联网云端处理图片，本地就能快速完成，隐私性也更强。未来智能手表、IoT芯片等资源受限的设备也能搭载更强大的视觉AI功能。

arXiv:2607.22564v1 Announce Type: new Abstract: Convolutional neural networks often contain redundant feature maps that increase storage and inference cost. This paper presents a loss-aware feature-map pruning framework using multi-armed bandits. Feature-map pruning is structured because it removes complete convolutional output channels and their producing filters rather than isolated scalar weights. Each candidate feature map is treated as an arm. At each play time, one map is temporarily masked and evaluated on a sampled mini-batch; the map is then restored and the observed loss change is converted into a safe-removal reward. After a fixed play budget, candidate maps are ranked by learned scores and the top-k maps are permanently removed with their filters, biases and corresponding next-layer input-channel kernels. The study evaluates UCB1 and Thompson Sampling, compares them with direct/oracle-style evaluation on LeNet/MNIST, and extends the evaluation to MNIST, CIFAR-10, CIFAR-100, SVHN, CUB-200-2011 and Oxford Flowers 102. Results show that UCB1 and Thompson Sampling preserve accuracy close to unpruned models while removing feature maps and reducing convolutional computation. Friedman and Nemenyi tests show that UCB1 obtains the highest mean rank, followed by Thompson Sampling; both significantly outperform greedy and magnitude-based pruning while remaining statistically comparable to the original unpruned model.