Filters








5,592 Hits in 4.9 sec

A Differentiable Ranking Metric Using Relaxed Sorting Operation for Top-K Recommender Systems [article]

Hyunsung Lee, Yeongjae Jang, Jaekwang Kim, Honguk Woo
2020 arXiv   pre-print
In this work, we present DRM (differentiable ranking metric) that mitigates the inconsistency and improves recommendation performance by employing the differentiable relaxation of ranking metrics.  ...  A recommender system generates personalized recommendations for a user by computing the preference score of items, sorting the items according to the score, and filtering top-K items with high scores.  ...  hence improve the performance of top-K recommendations. We show that the DRM objective is readily incorporated into the existing factor based recommenders via joint learning.  ... 
arXiv:2008.13141v4 fatcat:bgxf3itixjbypblycbng3nvnuq

PiRank: Scalable Learning To Rank via Differentiable Sorting [article]

Robin Swezey, Aditya Grover, Bruno Charron, Stefano Ermon
2021 arXiv   pre-print
We propose PiRank, a new class of differentiable surrogates for ranking, which employ a continuous, temperature-controlled relaxation to the sorting operator based on NeuralSort [1].  ...  This gap arises because ranking metrics typically involve a sorting operation which is not differentiable w.r.t. the model parameters.  ...  The k rows of P̂ are used as the top-k rows of the relaxed sorting operator Pbsort(ŷ) (τ ). This approach is equivalent to NeuralSort, yielding Eq. 12 for d = 1.  ... 
arXiv:2012.06731v2 fatcat:amtphimdijcodefysa67wxv2mu

Hashing as Tie-Aware Learning to Rank

Kun He, Fatih Cakir, Sarah Adel Bargal, Stan Sclaroff
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
Then, to optimize tie-aware ranking metrics, we derive their continuous relaxations, and perform gradient-based optimization with deep neural networks.  ...  In this paper, we develop learning to rank formulations for hashing, aimed at directly optimizing ranking-based evaluation metrics such as Average Precision (AP) and Normalized Discounted Cumulative Gain  ...  Acknowledgements The authors would like to thank Qinxun Bai, Peter Gacs, and Dora Erdos for helpful discussions.  ... 
doi:10.1109/cvpr.2018.00423 dblp:conf/cvpr/0003CBS18 fatcat:x23fwsjwofbg3kt44lxuustjkq

Hashing as Tie-Aware Learning to Rank [article]

Kun He, Fatih Cakir, Sarah Adel Bargal, Stan Sclaroff
2018 arXiv   pre-print
Then, to optimize tie-aware ranking metrics, we derive their continuous relaxations, and perform gradient-based optimization with deep neural networks.  ...  In this paper, we develop learning to rank formulations for hashing, aimed at directly optimizing ranking-based evaluation metrics such as Average Precision (AP) and Normalized Discounted Cumulative Gain  ...  Acknowledgements The authors would like to thank Qinxun Bai, Peter Gacs, and Dora Erdos for helpful discussions.  ... 
arXiv:1705.08562v4 fatcat:uzo3q6h2cranjlmhw7qle7kwlm

Monotonic Differentiable Sorting Networks [article]

Felix Petersen, Christian Borgelt, Hilde Kuehne, Oliver Deussen
2022 arXiv   pre-print
Differentiable sorting algorithms allow training with sorting and ranking supervision, where only the ordering or ranking of samples is known.  ...  To address this issue, we propose a novel relaxation of conditional swap operations that guarantees monotonicity in differentiable sorting networks.  ...  We specify all necessary hyperparameters for each experiment. We use the same model architectures as in previous works.  ... 
arXiv:2203.09630v1 fatcat:c3aopk3erzhlpbcivwucryrwwm

Neural Auction: End-to-End Learning of Auction Mechanisms for E-Commerce Advertising [article]

Xiangyu Liu, Chuan Yu, Zhilin Zhang, Zhenzhe Zheng, Yu Rong, Hongtao Lv, Da Huo, Yiqing Wang, Dagui Chen, Jian Xu, Fan Wu, Guihai Chen (+1 others)
2021 arXiv   pre-print
In this paper, we design Deep Neural Auctions (DNAs) to enable end-to-end auction learning by proposing a differentiable model to relax the discrete sorting operation, a key component in auctions.  ...  We optimize the performance metrics by developing deep models to efficiently extract contexts from auctions, providing rich features for auction design.  ...  The authors would like to thank Rui Du, Haiping Huang, Haiyang He and Guan Wang who did the really hard work for online system implementation.  ... 
arXiv:2106.03593v2 fatcat:7o6z4bq2gbh3fcan2nusgcnmae

Fairness-Aware Ranking in Search Recommendation Systems with Application to LinkedIn Talent Search [article]

Sahin Cem Geyik, Stuart Ambler, Krishnaram Kenthapadi
2019 arXiv   pre-print
We present a framework for quantifying and mitigating algorithmic bias in mechanisms designed for ranking individuals, typically used as part of web-scale search and recommendation systems.  ...  For a given search or recommendation task, our algorithms seek to achieve a desired distribution of top ranked results with respect to one or more protected attributes.  ...  Driscoll, Gurwinder Gulati, Rachel Kumar, Divyakumar Menghani, Chenhui Zhai, and Yani Zhang for working closely with us during the development of the product.  ... 
arXiv:1905.01989v2 fatcat:kvfbi37jlbeg7cxjimnrkr3spy

Differential Geometric Retrieval of Deep Features [article]

Y Qian and E Vazquez and B Sengupta
2017 arXiv   pre-print
In this paper, we compare distance metrics (and divergences) to rank features generated from a neural network, for content-based image retrieval.  ...  Comparing images to recommend items from an image-inventory is a subject of continued interest.  ...  Top left image is the query used.  ... 
arXiv:1702.06383v2 fatcat:w5hv5g72zbaltgngmwzconco5q

Towards Amortized Ranking-Critical Training for Collaborative Filtering [article]

Sam Lobel, Chunyuan Li, Jianfeng Gao, Lawrence Carin
2020 arXiv   pre-print
While simple and often effective, MLE-based training does not directly maximize the recommendation-quality metrics one typically cares about, such as top-N ranking.  ...  Collaborative filtering is widely used in modern recommender systems.  ...  One salient component of a ranking-based Oracle metric ω * is to sort π p . The sorting operation is nondifferentiable, rendering it impossible to directly use ω * as the critic.  ... 
arXiv:1906.04281v2 fatcat:fkjchyktvvghpegkqkx5ogcci4

COD: Iterative Utility Elicitation for Diversified Composite Recommendations

Khalid Alodhaibi, Alexander Brodsky, George A. Mihaila
2010 2010 43rd Hawaii International Conference on System Sciences  
This paper studies and proposes methods for providing recommendations on composite bundles of products and services that are dynamically defined using database views extended with decision optimization  ...  A framework is proposed for finding a diverse recommendation set when no prior knowledge on user preference is given.  ...  We proposed a framework for finding a diverse recommendation set, when no prior knowledge on user preference is given, which includes (1) finding recommendation cluster, (2) user utility elicitation using  ... 
doi:10.1109/hicss.2010.108 dblp:conf/hicss/AlodhaibiBM10 fatcat:7mihqf6xyngszacx7c2d7rpnym

Privacy in Search Logs [article]

Michaela Goetz, Ashwin Machanavajjhala, Guozhang Wang, Xiaokui Xiao, Johannes Gehrke
2011 arXiv   pre-print
We then demonstrate that the stronger guarantee ensured by ϵ-differential privacy unfortunately does not provide any utility for this problem.  ...  Our paper concludes with a large experimental study using real applications where we compare ZEALOUS and previous work that achieves k-anonymity in search log publishing.  ...  Acknowledgments We would like to thank our colleagues Filip Radlinski and Yisong Yue for helpful discussions about the usage of search logs.  ... 
arXiv:0904.0682v4 fatcat:etvjnfvpzfdd3gms34vk3gh6xu

Active top-K ranking from noisy comparisons

Soheil Mohajer, Changho Suh
2016 2016 54th Annual Allerton Conference on Communication, Control, and Computing (Allerton)  
., Strong Stochastic Transitivity model, BTL model and uniform noise model), we characterize upper bounds on the sample size required for top-K sorting as well as for top-K partitioning.  ...  We consider two settings: (1) top-K sorting in which the goal is to recover the top-K items in order out of n items; (2) top-K partitioning where only the set of top-K items is desired.  ...  Acknowledgment The authors would like to thank the reviewers who gave useful comments. C.  ... 
doi:10.1109/allerton.2016.7852326 dblp:conf/allerton/MohajerS16 fatcat:geq6mx7bvjeqpbtwgebso43gje

Evaluating Recommender System Stability with Influence-Guided Fuzzing

David Shriver, Sebastian Elbaum, Matthew B. Dwyer, David S. Rosenblum
2019 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
We implement our approach and evaluate it on several recommender algorithms using the MovieLens dataset.  ...  Since unstable recommendations can lead to distrust, loss of profits, and a poor user experience, it is important to test recommender system stability.  ...  Given D and a parameter k for computing top-k rankings, a recommender algorithm Q computes a partial function rank k,Q : I × U → [1, k] for k ≤ |I|, where rank k,Q (i, u) = ⊥ if item i is not ranked for  ... 
doi:10.1609/aaai.v33i01.33014934 fatcat:ucnq4w4jd5gzni3symk4woyv3u

Mostra: A Flexible Balancing Framework to Trade-off User, Artist and Platform Objectives for Music Sequencing [article]

Emanuele Bugliarello and Rishabh Mehrotra and James Kirk and Mounia Lalmas
2022 arXiv   pre-print
streaming platform, we present insights on the trade-offs that exist across different objectives, and demonstrate that the proposed framework leads to a superior, just-in-time balancing across the various metrics  ...  Top-left: gains in sessions with only Boost non-user objective. Top-centre: gains in sessions with more tracks for Boost, than for Exposure, than for Discovery; etc.  ...  Relevance ranker: Ranks tracks by estimating user-track relevance scores using the cosine similarity between contextual embeddings for each user and for each track.  ... 
arXiv:2204.10463v1 fatcat:wqgfxa56zfg3nmj2gh4fg5f6lu

Evaluating Stochastic Rankings with Expected Exposure [article]

Fernando Diaz and Bhaskar Mitra and Michael D. Ekstrand and Asia J. Biega and Ben Carterette
2020 arXiv   pre-print
We believe that measuring and optimizing expected exposure metrics using randomization opens a new area for retrieval algorithm development and progress.  ...  Importantly, this methodology relaxes classic information retrieval assumptions, allowing a system, in response to a query, to produce a distribution over rankings instead of a single fixed ranking.  ...  Figure 4 : 4 Sorting PL and RT runs by RBP. Table 1 : 1 Results for optimizing towards expected exposure and demographic parity using different ranking objectives.  ... 
arXiv:2004.13157v1 fatcat:hlj3mwt36fdivpkdp27b5nst2q
« Previous Showing results 1 — 15 out of 5,592 results