多智能体协作通信研究：谁该说话需成对测量
The Tower of Babel in artificial intelligence: who should speak, and when it pays to listen
arXiv 上这篇论文挺有意思：两个分类器怎么分工传消息，联合设计比各自设计高 18 到 22 个百分点，还给了什么时候该加搭档的判据。
arXiv 论文研究多个分类器各自观察物体不同部分、通过短消息协作识别类别的任务。在三个数字数据集上，联合设计的 2-bit 消息比分开设计的平均高 18.07 到 22.68 个百分点，且在全部 24 种配置中都有提升。固定约定（一方报告行、一方报告列）仅学习接收端即可恢复约一半增益。在 MNIST 上，6 个共享比特平均消除最强单发端 49.54% 的错误。研究还用可解模型证明单凭个体统计无法判断搭档是否有帮助，配对测量则正确预测了多数传感器团队会降低新参与者上的平均准确率。
The Tower of Babel in artificial intelligence: who should speak, and when it pays to listen
Two machines can each recognize an object well and still fail together. We study classifiers that see different parts of one object and send short messages to a receiver that identifies its class. Coordination asks how to design messages together; cooperation asks whether the team outperforms one sender given the whole budget. On three digit datasets, messages designed together beat messages designed apart by 18.07 to 22.68 percentage points on average with two-bit messages, with gains in all 24 configurations. A fixed agreement, you report the row and I the column, recovers about half this gain, with only the receiver learned. Senders that draw the same distinctions waste a narrow vocabulary, and senders whose distinctions cross multiply it. A team has no more possible messages than one sender with the same budget, so a partner must bring evidence the first lacks. A solvable model shows individual statistics alone cannot determine whether a partnership helps. On MNIST digits, six shared bits remove 49.54 percent of the strongest tested solo's errors on average. Paired measurements correctly predicted most sensor teams would reduce average accuracy on new participants. Who should speak is a property of the pair, and must be measured in pairs.
