Predicting Query-Item Relationship using Adversarial Training and Robust Modeling Techniques
This work addresses a challenging problem in search and recommendation systems, but it is incremental as it builds on existing methods with robustness enhancements.
The paper tackled the problem of predicting search query-item relationships, particularly for unseen queries, by combining pre-trained transformers and LSTMs with adversarial training and other robustness techniques, achieving 10th place in the KDD Cup 2022 Product Substitution Classification task.
We present an effective way to predict search query-item relationship. We combine pre-trained transformer and LSTM models, and increase model robustness using adversarial training, exponential moving average, multi-sampled dropout, and diversity based ensemble, to tackle an extremely difficult problem of predicting against queries not seen before. All of our strategies focus on increasing robustness of deep learning models and are applicable in any task where deep learning models are used. Applying our strategies, we achieved 10th place in KDD Cup 2022 Product Substitution Classification task.