Recent work has shown that a much simpler model, simple graph convolution (SGC) (Wu et al., 2019),iscompetitivewithGCNs incommon graph machine learning benchmarks.

In addition, we present an alternative design that permits a near-optimal sublinear decoding timeofO(klog2k+klogn).

We also provide examples where biased estimators outperform their unbiased counterparts or where unbiased versions are simply not available. Finally, we demonstrate the effectiveness of our framework through experimental results that validate our theoretical findings.

Neural Information Processing Systems http://nips.cc/

However, a naive scheme to train such amodel results inadegenerate solution.