Industrial purposes of machine studying are generally composed of assorted objects which have differing information modalities or characteristic distributions. Heterogeneous graphs (HGs) supply a unified view of those multimodal information programs by defining a number of sorts of nodes (for every information kind) and edges (for the relation between information objects). As an illustration, e-commerce networks might need [user, product, review] nodes or video platforms might need [channel, user, video, comment] nodes. Heterogeneous graph neural networks (HGNNs) study node embeddings summarizing every node’s relationships right into a vector. Nonetheless, in actual world HGs, there’s usually a label imbalance difficulty between totally different node varieties. Because of this label-scarce node varieties can not exploit HGNNs, which hampers the broader applicability of HGNNs.
In “Zero-shot Switch Studying inside a Heterogeneous Graph through Information Switch Networks”, offered at NeurIPS 2022, we suggest a mannequin known as a Information Switch Community (KTN), which transfers data from label-abundant node varieties to zero-labeled node varieties utilizing the wealthy relational data given in a HG. We describe how we pre-train a HGNN mannequin with out the necessity for fine-tuning. KTNs outperform state-of-the-art switch studying baselines by as much as 140% on zero-shot studying duties, and can be utilized to enhance many current HGNN fashions on these duties by 24% (or extra).
|KTNs rework labels from one kind of knowledge (squares) by way of a graph to a different kind (stars).|
What’s a heterogeneous graph?
A HG consists of a number of node and edge varieties. The determine under exhibits an e-commerce community offered as a HG. In e-commerce, “customers” buy “merchandise” and write “evaluations”. A HG presents this ecosystem utilizing three node varieties [user, product, review] and three edge varieties [user-buy-product, user-write-review, review-on-product]. Particular person merchandise, customers, and evaluations are then offered as nodes and their relationships as edges within the HG with the corresponding node and edge varieties.
|E-commerce heterogeneous graph.|
Along with all connectivity data, HGs are generally given with enter node attributes that summarize every node’s data. Enter node attributes might have totally different modalities throughout totally different node varieties. As an illustration, pictures of merchandise may very well be given as enter node attributes for the product nodes, whereas textual content may be given as enter attributes to assessment nodes. Node labels (e.g., the class of every product or the class that the majority pursuits every person) are what we wish to predict on every node.
HGNNs and label shortage points
HGNNs compute node embeddings that summarize every node’s native buildings (together with the node and its neighbor’s data). These node embeddings are utilized by a classifier to foretell every node’s label. To coach a HGNN mannequin and a classifier to foretell labels for a selected node kind, we require quantity of labels for the kind.
A standard difficulty in industrial purposes of deep studying is label shortage, and with their numerous node varieties, HGNNs are much more more likely to face this problem. As an illustration, publicly out there content material node varieties (e.g., product nodes) are abundantly labeled, whereas labels for person or account nodes might not be out there because of privateness restrictions. Because of this in most traditional coaching settings, HGNN fashions can solely study to make good inferences for a number of label-abundant node varieties and might normally not make any inferences for any remaining node varieties (given the absence of any labels for them).
Switch studying on heterogeneous graphs
Zero-shot switch studying is a method used to enhance the efficiency of a mannequin on a goal area with no labels by utilizing the data discovered by the mannequin from one other associated supply area with adequately labeled information. To use switch studying to resolve this label shortage difficulty for sure node varieties in HGs, the goal area could be the zero-labeled node varieties. Then what could be the supply area? Earlier work generally units the supply area as the identical kind of nodes situated in a distinct HG, assuming these nodes are abundantly labeled. This graph-to-graph switch studying method pre-trains a HGNN mannequin on the exterior HG after which runs the mannequin on the unique (label-scarce) HG.
Nonetheless, these approaches should not relevant in lots of real-world eventualities for 3 causes. First, any exterior HG that may very well be utilized in a graph-to-graph switch studying setting would virtually absolutely be proprietary, thus, possible unavailable. Second, even when practitioners might receive entry to an exterior HG, it’s unlikely the distribution of that supply HG would match their goal HG nicely sufficient to use switch studying. Lastly, node varieties affected by label shortage are more likely to endure the identical difficulty on different HGs (e.g., privateness points on person nodes).
Our method: Switch studying between node varieties inside a heterogeneous graph
Right here, we make clear a extra sensible supply area, different node varieties with plentiful labels situated on the identical HG. As a substitute of utilizing further HGs, we switch data inside a single HG (assumed to be absolutely owned by the practitioners) throughout several types of nodes. Extra particularly, we pre-train a HGNN mannequin and a classifier on a label-abundant (supply) node kind, then reuse the fashions on the zero-labeled (goal) node varieties situated in the identical HG with out extra fine-tuning. The one requirement is that the supply and goal node varieties share the identical label set (e.g., within the e-commerce HG, product nodes have a label set describing product classes, and person nodes share the identical label set describing their favourite purchasing classes).
Why is it difficult?
Sadly, we can not straight reuse the pre-trained HGNN and classifier on the goal node kind. One essential attribute of HGNN architectures is that they’re composed of modules specialised to every node kind to completely study the multiplicity of HGs. HGNNs use distinct units of modules to compute embeddings for every node kind. Within the determine under, blue- and red-colored modules are used to compute node embeddings for the supply and goal node varieties, respectively.
|HGNNs are composed of modules specialised to every node kind and use distinct units of modules to compute embeddings of various node varieties. Extra particulars may be discovered within the paper.|
Whereas pre-training HGNNs on the supply node kind, source-specific modules within the HGNNs are nicely educated, nevertheless target-specific modules are under-trained as they’ve solely a small quantity of gradients flowing into them. That is proven under, the place we see that the L2 norm of gradients for goal node varieties (i.e., Mtt) are a lot decrease than for supply varieties (i.e., Mss). On this case a HGNN mannequin outputs poor node embeddings for the goal node kind, which ends up in poor process efficiency.
|In HGNNs, goal type-specific modules obtain zero or solely a small quantity of gradients throughout pre-training on the supply node kind, resulting in poor efficiency on the goal node kind.|
KTN: Trainable cross-type switch studying for HGNNs
Our work focuses on remodeling the (poor) goal node embeddings computed by a pre-trained HGNN mannequin to comply with the distribution of the supply node embeddings. Then the classifier, pre-trained on the supply node kind, may be reused for the goal node kind. How can we map the goal node embeddings to the supply area? To reply this query, we examine how HGNNs compute node embeddings to study the connection between supply and goal distributions.
HGNNs mixture linked node embeddings to enhance a goal node’s embeddings in every layer. In different phrases, the node embeddings for each supply and goal node varieties are up to date utilizing the identical enter — the earlier layer’s node embeddings of any linked node varieties. Because of this they are often represented by one another. We show this relationship theoretically and discover there’s a mapping matrix (outlined by HGNN parameters) from the goal area to the supply area (extra particulars in Theorem 1 within the paper). Based mostly on this theorem, we introduce an auxiliary neural community, which we confer with as a Information Switch Community (KTN), that receives the goal node embeddings after which transforms them by multiplying them with a (trainable) mapping matrix. We then outline a regularizer that’s minimized together with the efficiency loss within the pre-training section to coach the KTN. At take a look at time, we map the goal embeddings computed from the pre-trained HGNN to the supply area utilizing the educated KTN for classification.
|In HGNNs, the ultimate node embeddings of each supply and goal varieties are computed from totally different mathematical capabilities (f(): supply, g(): goal) which use the identical enter — the earlier layer’s node embeddings.|
To look at the effectiveness of KTNs, we ran 18 totally different zero-shot switch studying duties on two public heterogeneous graphs, Open Educational Graph and Pubmed. We examine KTN with eight state-of-the-art switch studying strategies (DAN, JAN, DANN, CDAN, CDAN-E, WDGRL, LP, EP). Proven under, KTN constantly outperforms all baselines on all duties, beating switch studying baselines by as much as 140% (as measured by Normalized Discounted Cumulative Achieve, a rating metric).
|Zero-shot switch studying on Open Educational Graph (OAG-CS) and Pubmed datasets. The colours characterize totally different classes of switch studying baselines towards which the outcomes are in contrast. Yellow: Use statistical properties (e.g., imply, variance) of distributions. Inexperienced: Use adversarial fashions to switch data. Orange: Switch data straight through graph construction utilizing label propagation.|
Most significantly, KTN may be utilized to virtually all HGNN fashions which have node and edge type-specific parameters and enhance their zero-shot efficiency on track domains. As proven under, KTN improves accuracy on zero-labeled node varieties throughout six totally different HGNN fashions(R-GCN, HAN, HGT, MAGNN, MPNN, H-MPNN) by as much as 190%.
|KTN may be utilized to 6 totally different HGNN fashions and enhance their zero-shot efficiency on track domains.|
Varied ecosystems in business may be offered as heterogeneous graphs. HGNNs summarize heterogeneous graph data into efficient representations. Nonetheless, label shortage points on sure sorts of nodes stop the broader software of HGNNs. On this publish, we launched KTN, the primary cross-type switch studying methodology designed for HGNNs. With KTN, we are able to absolutely exploit the richness of heterogeneous graphs through HGNNs no matter label shortage. See the paper for extra particulars.
This paper is joint work with our co-authors John Palowitch (Google Analysis), Dustin Zelle (Google Analysis), Ziniu Hu (Intern, Google Analysis), and Russ Salakhutdinov (CMU). We thank Tom Small for creating the animated determine on this weblog publish.