发表机构
University of Johannesburg（约翰内斯堡大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出结构化亲和性等关键机制，构建无需重放、反向传播的深度人工免疫网络，在多视觉数据集上实现高效无监督视觉类增量记忆学习。
AI 中文摘要
人工免疫网络（AINs）是天然的记忆形成系统，但传统视觉AINs常依赖忽略空间结构的扁平化向量亲和性。本文研究无梯度的结构化免疫亲和性能否使深度AINs成为无需重放的视觉类增量表示记忆学习器。视觉B细胞被形式化为结构化模板，包括移位模板亲和性、零归一化互相关（ZNCC）滤波器及特征图绑定轮廓。免疫库兼具记忆和表示诱导基的作用，深度则通过将绑定轮廓响应图传递至后续免疫层实现。所得深度AIN表现出自适应潜坐标重组：随着新类别的加入，绑定轮廓空间会演化，同时保留早期类别的可恢复结构。在sklearn digits、MNIST、Fashion-MNIST和KMNIST上的实验表明，保留响应图至关重要；标量绑定轮廓变体表现不佳，而特征图深度AIN无需重放、标签驱动的免疫更新或通过免疫层的反向传播，即可学习具有类别区分性的视觉记忆。在sklearn digits上，遇到全部10个类别后，基于学习到的绑定轮廓拟合的下游探针，使用逻辑回归达到0.939的最终平衡准确率，使用1近邻达到0.902，初始类别保留率为0.978。自适应分层尺度校准进一步将两层特征图深度AIN的平衡准确率提升至0.978；采用相同校准规则，Fashion-MNIST达到0.814，KMNIST达到0.853。这些探针是外部验证工具，而非AIN的组成部分。研究结果确定结构化亲和性、响应图保留、自适应潜重组及分层尺度校准是无重放视觉免疫记忆的关键机制。
英文摘要
Artificial immune networks (AINs) are naturally memory-forming systems, but conventional visual AINs often rely on flattened vector affinity that ignores spatial structure. This paper studies whether structured, gradient-free immune affinity can make Deep AINs viable as replay-free visual class-incremental representation-memory learners. Visual B-cells are formalized as structured templates, including shifted-template affinity, zero-normalized cross-correlation (ZNCC) filters, and feature-map binding profiles. A repertoire is treated both as memory and as a representation-inducing basis, while depth is obtained by passing binding-profile response maps to subsequent immune layers. The resulting Deep AIN exhibits adaptive latent coordinate reorganization: as new classes arrive, the binding-profile space evolves while retaining recoverable structure for earlier classes. Experiments on sklearn digits, MNIST, Fashion-MNIST, and KMNIST show that preserving response maps is critical. Scalar binding-profile variants underperform, whereas feature-map Deep AINs learn class-discriminative visual memory without replay, label-driven immune updates, or backpropagation through the immune layers. On sklearn digits, downstream probes fitted on the learned binding profiles reach 0.939 final balanced accuracy with logistic regression and 0.902 with 1-nearest-neighbour after all ten classes are encountered, with initial-class retention of 0.978. Adaptive layer-wise scale calibration further improves the two-layer feature-map Deep AIN to 0.978 balanced accuracy. With the same calibration rule, Fashion-MNIST reaches 0.814 and KMNIST reaches 0.853. These probes are external validation tools, not components of the AIN. The results identify structured affinity, response-map preservation, adaptive latent reorganization, and layer-wise scale calibration as key mechanisms for replay-free visual immune memory.
Comments18 pages, 3 figures
