通过缩小全连接层，让卷积网络学会不依赖词汇的发音规则。
A Technique for Isolating Lexically-Independent Phonetic Dependencies in Generative CNNs
- 压缩全连接层至8通道，迫使网络依赖卷积层捕捉发音规律。
- 绕过全连接层输入随机特征图，仍能生成符合音系规则的音频。
- 揭示卷积层可动态泛化词汇无关的发音约束，适合语音生成研究者。
深度神经网络能否从词汇学习中表征音系规律仍是未解问题。本研究（1）考察在原始语音波形上训练的生成卷积神经网络（CNN）的词汇无关泛化能力，（2）探索将全连接层瓶颈从1024通道缩减至8通道的影响。最终提出一种新探查技术：在窄全连接瓶颈下，绕过全连接层，向卷积块输入随机特征图生成音频。结果显示，此类输出与经全连接层生成的输出一样，均受训练中音系限制的相同偏差影响。这表明卷积层可在词汇约束之外动态泛化语音依赖关系。
原文摘要 · Abstract (English)
The ability of deep neural networks (DNNs) to represent phonotactic generalizations derived from lexical learning remains an open question. This study (1) investigates the lexically-invariant generalization capacity of generative convolutional neural networks (CNNs) trained on raw audio waveforms of lexical items and (2) explores the consequences of shrinking the fully-connected layer (FC) bottleneck from 1024 channels to 8 before training. Ultimately, a novel technique for probing a model's lexically-independent generalizations is proposed that works only under the narrow FC bottleneck: generating audio outputs by bypassing the FC and inputting randomized feature maps into the convolutional block. These outputs are equally biased by a phonotactic restriction in training as are outputs generated with the FC. This result shows that the convolutional layers can dynamically generalize phonetic dependencies beyond lexically-constrained configurations learned by the FC.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。