Search for: All records

There has been significant progress in improving the performance of graph neural networks (GNNs) through enhancements in graph data, model architecture design, and training strategies. For fairness in graphs, recent studies achieve fair representations and predictions through either graph data pre-processing (e.g., node feature masking, and topology rewiring) or fair training strategies (e.g., regularization, adversarial debiasing, and fair contrastive learning). How to achieve fairness in graphs from the model architecture perspective is less explored. More importantly, GNNs exhibit worse fairness performance compared to multilayer perception since their model architecture (i.e., neighbor aggregation) amplifies biases. To this end, we aim to achieve fairness via a new GNN architecture. We propose Fair Message Passing (FMP) designed within a unified optimization framework for GNNs. Notably, FMP explicitly renders sensitive attribute usage in forward propagation for node classification task using cross-entropy loss without data pre-processing. In FMP, the aggregation is first adopted to utilize neighbors' information and then the bias mitigation step explicitly pushes demographic group node presentation centers together.In this way, FMP scheme can aggregate useful information from neighbors and mitigate bias to achieve better fairness and prediction tradeoff performance. Experiments on node classification tasks demonstrate that the proposed FMP outperforms several baselines in terms of fairness and accuracy on three real-world datasets. The code is available at https://github.com/zhimengj0326/FMP.

 

Embedding is widely used in recommendation models to learn feature representations. However, the traditional embedding technique that assigns a fixed size to all categorical features may be suboptimal due to the following reasons. In recommendation domain, the majority of categorical features' embeddings can be trained with less capacity without impacting model performance, thereby storing embeddings with equal length may incur unnecessary memory usage. Existing work that tries to allocate customized sizes for each feature usually either simply scales the embedding size with feature's popularity or formulates this size allocation problem as an architecture selection problem. Unfortunately, most of these methods either have large performance drop or incur significant extra time cost for searching proper embedding sizes. In this article, instead of formulating the size allocation problem as an architecture selection problem, we approach the problem from a pruning perspective and proposePruning-basedMulti-sizeEmbedding (PME) framework. During the search phase, we prune the dimensions that have the least impact on model performance in the embedding to reduce its capacity. Then, we show that the customized size of each token can be obtained by transferring the capacity of its pruned embedding with significant less search cost. Experimental results validate that PME can efficiently find proper sizes and hence achieve strong performance while significantly reducing the number of parameters in the embedding layer.

 

It has been suggested that a cellularly unstable laminar flame, which is freely propagating in unbounded space, can accelerate and evolve into a turbulent flame with the neighbouring flow exhibiting the basic characteristics of turbulence. Famously known as self-turbulization , this conceptual transition in the flow regime, which arises from local interactions between the propagating wrinkled flamefront and the flow, is critical in extreme events such as the deflagration-to-detonation transition (DDT) leading to supernova explosions. Recognizing that such a transition in the flow regime has not been conclusively demonstrated through experiments, in this work, we present experimental measurements of flow characteristics of flame-generated ‘turbulence’ for expanding cellular laminar flames. The energy spectra of such ‘turbulence’ at different stages of cellular instability are analysed. A subsequent scaling analysis points out that the observed energy spectra are driven by the fractal topology of the cellularly unstable flamefront. 

Sulfur compounds are an important constituent of particulate matter, with impacts on climate and public health. While most sulfur observed in particulate matter has been assumed to be sulfate, laboratory experiments reveal that hydroxymethanesulfonate (HMS), an adduct formed by aqueous phase chemical reaction of dissolved HCHO and SO₂, may be easily misinterpreted in measurements as sulfate. Here we present observational and modeling evidence for a ubiquitous global presence of HMS. We find that filter samples collected in Shijiazhuang, China, and examined with ion chromatography within 9 days show as much as 7.6 μg m⁻³of HMS, while samples from Singapore examined 9–18 months after collection reveal ~0.6 μg m⁻³of HMS. The Shijiazhuang samples show only minor traces of HMS 4 months later, suggesting that HMS had decomposed over time during sample storage. In contrast, the Singapore samples do not clearly show a decline in HMS concentration over 2 months of monitoring. Measurements from over 150 sites, primarily derived from the IMPROVE network across the United States, suggest the ubiquitous presence of HMS in at least trace amounts as much as 60 days after collection. The degree of possible HMS decomposition in the IMPROVE observations is unknown. Using the GEOS‐Chem chemical transport model, we estimate that HMS may account for 10% of global particulate sulfur in continental surface air and over 25% in many polluted regions. Our results suggest that reducing emissions of HCHO and other volatile organic compounds may have a co‐benefit of decreasing particulate sulfur.