Session: Models and Learning II

Date: Wednesday September 21, 11:00 AM – 12:30 PM (PDT)

  • PAMARRS: A Framework for multi-objective risk-aware route recommendation using Multitask-Transformer
    by Bhumika . (IIT Jodhpur, India), Debasis Das (Indian Institute of Technology (IIT), India)

    One of the most significant map services in navigation applications is route recommendation. However, most route recommendation systems only recommend trips based on time and distance, impacting quality-of-experience (QoE) and route selection. This paper introduces a novel framework, namely MARRS, a multi-objective route recommendation system based on heterogeneous urban sensing open data (i.e., crime, accident, traffic flow, road network, meteorological, calendar event and point of interest distributions). We introduce a wide, deep and multitask-learning (WD-MTL) framework that uses transformer to extract spatial, temporal, and semantic correlation for predicting crime, accident, and traffic flow of particular road segment. Later, for a particular source and destination, adaptive epsilon constraint technique is used to optimized route satisfying multiple objective functions. The experimental results demonstrate the feasibility of figuring out the safest and efficient route selection.

    Full text in ACM Digital Library

  • PAAdversary or Friend? An adversarial Approach to Improving Recommender Systems
    by Pannaga Shivaswamy (Netflix Inc, United States), Dario Garcia Garcia (Netflix, United States)

    Typical recommender systems models are trained to have good average performance across all users or items. In practice, this results in model performance that is good for some users but sub-optimal for many users. In this work, we consider adversarially trained machine learning models and extend them to recommender systems problems. The adversarial models are trained with no additional demographic or other information than already available to the learning algorithm. We show that adversarially reweighted learning models give more emphasis to dense areas of the feature-space that incur high loss during training. We show that a straightforward adversarial model can fail and that a carefully designed adversarial model can perform well. The proposed models are trained using a standard gradient descent/ascent approach that can be easily adapted to many recommender problems. We compare our results with an inverse propensity weighting based baseline that also works well in practice. We delve deep into the underlying experimental results and show that, for the users who are under-served by the baseline model, the adversarial models can achieve significantly better results.

    Full text in ACM Digital Library

  • PADual Attentional Higher Order Factorization Machines
    by Arindam Sarkar (Amazon, India), Dipankar Das (Amazon, India), Vivek Sembium (Amazon, India), Prakash Mandayam Comar (Amazon, India)

    Numerous problems of practical significance such as click-through rate (CTR) prediction, forecasting, tagging and so on, involve complex interaction of various user, item and context features. Manual feature engineering has been used in the past to model these combinatorial features but it requires domain expertise and becomes prohibitively expensive as number of features increases. Feedforward neural networks alleviate the need for manual feature engineering to a large extent and have shown impressive performance across multiple domains due to their ability to learn arbitrary functions. Despite multiple layers of non-linear projections, neural networks are limited in their ability to efficiently model functions with higher order interaction terms. In recent years, Factorization Machines and its variants have been proposed to explicitly capture higher order combinatorial interactions. However not all feature interactions are equally important, and in sparse data settings, without a suitable suppression mechanism, this might result into noisy terms during inference and hurt model generalization. In this work we present Dual Attentional Higher Order Factorization Machine (DA-HoFM), a unified attentional higher order factorization machine which leverages a compositional architecture to compute higher order terms with complexity linear in terms of maximum interaction degree. Equipped with \textit{sparse} dual \textit{attention} mechanism, DA-HoFM summarizes interaction terms at each layer, and is able to efficiently select important higher order terms.
    We empirically demonstrate effectiveness of our proposed models on the task of CTR prediction, where our model exhibits superior performance compared to the recent \textit{state-of-the-art} models, outperforming them by up to 6.7% on the logloss metric.

    Full text in ACM Digital Library

Diamond Supporter
Platinum Supporters
Gold Supporters
Challenge Sponsor
Special Supporters