Feature attribution methods are widely used to explain neural models by
determining the influence of individual input features on the models' outputs.
We propose a novel feature attribution method, CAFE (Conflict-Aware
Feature-wise Explanations), that addresses three limitations of the existing
methods: their disregard for the impact of conflicting features, their lack of
consideration for the influence of bias terms, and an overly high sensitivity
to local variations in the underpinning activation functions. Unlike other
methods, CAFE provides safeguards against overestimating the effects of neuron
inputs and separately traces positive and negative influences of input features
and biases, resulting in enhanced robustness and increased ability to surface
feature conflicts. We show experimentally that CAFE is better able to identify
conflicting features on synthetic tabular data and exhibits the best overall
fidelity on several real-world tabular datasets, while being highly
computationally efficient.

我们提出了一种新颖的特征归因方法 CAFE（冲突感知特征解释），它解决了现有方法中的三个限制：它们忽视冲突特征的影响，不考虑偏差项的影响，以及对基础激活函数的局部变化过于敏感。与其他方法不同，CAFE 在溯源神经元输入的影响上提供了防止效应被高估的保护措施，并单独跟踪输入特征和偏差的正负影响，从而增强了鲁棒性，并增加了表面冲突特征的能力。我们的实验证明，CAFE 能够更好地识别合成表格数据中的冲突特征，并在几个真实世界的表格数据集上具有最好的整体保真度，同时具有很高的计算效率。

冲突感知特征解释

CAFE: Conflict-Aware Feature-wise Explanations

Dataset condensation aims at reducing the network training effort through
condensing a cumbersome training set into a compact synthetic one.
State-of-the-art approaches largely rely on learning the synthetic data by
matching the gradients between the real and synthetic data batches. Despite the
intuitive motivation and promising results, such gradient-based methods, by
nature, easily overfit to a biased set of samples that produce dominant
gradients, and thus lack global supervision of data distribution. In this
paper, we propose a novel scheme to Condense dataset by Aligning FEatures
(CAFE), which explicitly attempts to preserve the real-feature distribution as
well as the discriminant power of the resulting synthetic set, lending itself
to strong generalization capability to various architectures. At the heart of
our approach is an effective strategy to align features from the real and
synthetic data across various scales, while accounting for the classification
of real samples. Our scheme is further backed up by a novel dynamic bi-level
optimization, which adaptively adjusts parameter updates to prevent
over-/under-fitting. We validate the proposed CAFE across various datasets, and
demonstrate that it generally outperforms the state of the art: on the SVHN
dataset, for example, the performance gain is up to 11%. Extensive experiments
and analyses verify the effectiveness and necessity of proposed designs.

本文提出了一种名为 CAFE 的新方法来压缩数据集并 align 真实数据和合成数据的特征，以防止梯度偏差。我们的方案通过动态双级优化和适应性参数更新以防止过度拟合，并通过广泛实验验证了其有效性。