Filters








1,517 Hits in 4.7 sec

Learning with dataset bias in latent subcategory models

Dimitris Stamos, Samuele Martelli, Moin Nabi, Andrew McDonald, Vittorio Murino, Massimiliano Pontil
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
In this paper, we present a model which jointly learns an LSM for each dataset as well as a compound LSM.  ...  Latent subcategory models (LSMs) offer significant improvements over training linear support vector machines (SVMs).  ...  Discussion and Conclusion We presented a method for learning latent subcategories in presence of multiple biased datasets.  ... 
doi:10.1109/cvpr.2015.7298988 dblp:conf/cvpr/StamosMNMMP15 fatcat:kkzmm6ehlnhjvhvuh7ygicb23m

Mid-level Representation for Visual Recognition [article]

Moin Nabi
2015 arXiv   pre-print
We, additionally, study the outcomes provided by employing the subcategory-based models for undoing dataset bias.  ...  We specifically employ the discriminative patches in a subcategory-aware webly-supervised fashion.  ...  with Dataset Bias in Latent Subcategory Models”, D.  ... 
arXiv:1512.07314v1 fatcat:knmhkwxqk5aczis7ce6g2sv2wm

Structural Analysis of User Choices for Mobile App Recommendation [article]

Bin Liu, Yao Wu, Neil Zhenqiang Gong, Junjie Wu, Hui Xiong, Martin Ester
2016 arXiv   pre-print
Moreover, we design an efficient learning algorithm to estimate the parameters for the SUCM model.  ...  First, app markets have organized apps in a hierarchical taxonomy. Second, apps with similar functionalities are competing with each other.  ...  Thus, we associate each internal node z with a latent variable qz to represent the category/subcategory level properties, and we model the latent variable qz as a function of the latent variable of z's  ... 
arXiv:1605.07980v1 fatcat:ygar37scgrczlpg74wnut74pwu

How important are Deformable Parts in the Deformable Parts Model? [article]

Santosh K. Divvala and Alexei A. Efros and Martial Hebert
2012 arXiv   pre-print
A secondary contribution is the latent discriminative learning. Tertiary is the use of multiple components.  ...  The main stated contribution of the Deformable Parts Model (DPM) detector of Felzenszwalb et al.  ...  In terms of model learning, the DPM detector has the subcategory, as well as the part deformation parameters (four) as latent variables (for each of the 48 parts), while the visual subcategory detector  ... 
arXiv:1206.3714v1 fatcat:hy7gm6zyube3hnlumtfzds7m44

Refining Image Categorization by Exploiting Web Images and General Corpus [article]

Yazhou Yao, Jian Zhang, Fumin Shen, Xiansheng Hua, Wankou Yang and Zhenmin Tang
2017 arXiv   pre-print
Studies show that refining real-world categories into semantic subcategories contributes to better image modeling and classification.  ...  The following two major challenges are well studied: 1) noise in the labels of subcategories derived from the general corpus; 2) noise in the labels of images retrieved from the web.  ...  In [31] , both of positive and negative images are used to learn subcategories. Particularly, a new model by joint clustering and classification was proposed for discriminative sub-categorization.  ... 
arXiv:1703.05451v1 fatcat:gjpbnhlog5gq5j4n25iqvrvedy

Visually-Aware Personalized Recommendation using Interpretable Image Representations [article]

Charles Packer, Julian McAuley, Arnau Ramisa
2018 arXiv   pre-print
By using interpretable image representations generated with a unique feature learning process, our model learns to explain users' prior feedback in terms of their affinity towards specific visual attributes  ...  In this paper we propose a novel approach to personalized clothing recommendation that models the dynamics of individual users' visual preferences.  ...  In our case, the preference predictor extends the basic latent factor model by learning set of visual factors (θ u , θ i ), where θ T u θ i encodes a separate visual-speci c compatibility.  ... 
arXiv:1806.09820v2 fatcat:fau363r73vg3rjyzpqofgyazsy

Ups and Downs

Ruining He, Julian McAuley
2016 Proceedings of the 25th International Conference on World Wide Web - WWW '16  
In certain domains, such as fashion, modeling such preferences can be incredibly difficult, due to the need to simultaneously model the visual appearance of products as well as their evolution over time  ...  In this paper we build novel models for the One-Class Collaborative Filtering setting, where our goal is to estimate users' fashion-aware personalized ranking functions based on their past feedback.  ...  Our choice is to replace the stationary item bias term βi in Eq. 7 with a temporal counterpart βi(t) [19] . Per-Subcategory Temporal Dynamics.  ... 
doi:10.1145/2872427.2883037 dblp:conf/www/HeM16 fatcat:7rfv2qfmkjcltiybs7l3xybu74

Learning Improved Representations by Transferring Incomplete Evidence Across Heterogeneous Tasks [article]

Athanasios Davvetas, Iraklis A. Klampanos
2019 arXiv   pre-print
However, they often introduce noise or uncertainty in the labelset, that leads to decreased performance of supervised deep learning methods.  ...  In this paper we evaluate the effectiveness of a representation learning method that uses external categorical evidence called "Evidence Transfer", against low amount of corresponding evidence termed as  ...  The autoencoder is trained as a generative model that learns latent representation which approximate the data generation distribution.  ... 
arXiv:1912.10490v1 fatcat:fa2zh3mwnngnzdxf3yiiuggzve

Exploiting POI-Specific Geographical Influence for Point-of-Interest Recommendation

Hao Wang, Huawei Shen, Wentao Ouyang, Xueqi Cheng
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
In this paper, we exploit POI-specific geographical influence to improve POI recommendation.  ...  Experimental results on two real-world datasets demonstrate that POI-specific geographical influence significantly improves the performance of POI recommendation.  ...  Each POI in both datasets is associated with its longitude and latitude. Additionally, in the Foursquare dataset, each POI is marked by 8 categories and 240 subcategories.  ... 
doi:10.24963/ijcai.2018/539 dblp:conf/ijcai/WangSOC18 fatcat:b5li2wpxyrf7bkmp5aw5lae2li

Zero-Shot Recommender Systems [article]

Hao Ding, Yifei Ma, Anoop Deoras, Yuyang Wang, Hao Wang
2021 arXiv   pre-print
On the other hand, zero-shot learning promises some degree of generalization from an old dataset to an entirely new dataset. In this paper, we explore the possibility of zero-shot learning in RS.  ...  In terms of users, ZESRec builds upon recent advances on sequential RS to represent users using their interactions with items, thereby generalizing to unseen users as well.  ...  This reflects strong recency bias in news recommendation. In contrast, our zero-shot learning naturally comes with an inductive bias to only model the transition from user history to the next item.  ... 
arXiv:2105.08318v2 fatcat:tbpnyiaefrhnlnlwsjqj3qlmbu

Metadata Embeddings for User and Item Cold-start Recommendations [article]

Maciej Kula
2015 arXiv   pre-print
I present a hybrid matrix factorisation model representing users and items as linear combinations of their content features' latent factors.  ...  The model outperforms both collaborative and content-based models in cold-start or sparse interaction data scenarios (using both user and item metadata), and performs at least as well as a pure collaborative  ...  I train the LightFM models using stochastic gradient descent with an initial learning rate of 0.05. The latent dimensionality of the models is set to 64 for all models and experiments.  ... 
arXiv:1507.08439v1 fatcat:4thubzvzzbbbfn2t4fkkg2vp5y

Deep Belief Nets for Topic Modeling [article]

Lars Maaloe and Morten Arngren and Ole Winther
2015 arXiv   pre-print
This article also comes with a newly developed deep belief nets toolbox for topic modeling tailored towards performance evaluation of the DBN model and comparisons to the LDA model.  ...  Efficient retrieval can be carried out in the latent representation. We work both on public benchmarks and digital media content provided by Issuu, an online publishing platform.  ...  As mentioned, Issuu has applied labels to the dataset from the results of their LDA model with a 150-dimensional latent representation.  ... 
arXiv:1501.04325v1 fatcat:symffahs5ze4bf6qwmkdmyynhi

Quantifying Availability and Discovery in Recommender Systems via Stochastic Reachability [article]

Mihaela Curmei, Sarah Dean, Benjamin Recht
2021 arXiv   pre-print
Stochastic reachability can be used to detect biases in the availability of content and diagnose limitations in the opportunities for discovery granted to users.  ...  In this work, we consider how preference models in interactive recommendation systems determine the availability of content and users' opportunities for discovery.  ...  Acknowledgements This research is generously supported in part by ONR awards N00014-20-1-2497 and N00014-18-1-2833, NSF CPS award 1931853, and the DARPA Assured Autonomy program (FA8750-18-C-0101).  ... 
arXiv:2107.00833v1 fatcat:2puclhtbonccvhxeijzrd45fu4

Aspect-driven User Preference and News Representation Learning for News Recommendation [article]

Rongyao Wang, Wenpeng Lu, Shoujin Wang, Xueping Peng, Hao Wu, Qian Zhang
2022 arXiv   pre-print
Extensive experiments are done on the commonly used real-world dataset MIND, which demonstrate the superiority of our method compared with representative and state-of-the-art methods.  ...  In ANRS, news aspect-level encoder and user aspect-level encoder are devised to learn the fine-grained aspect-level representations of user's preferences and news characteristics respectively, which are  ...  Second, the single neural network, i.e., CNN only achieves similar performance with the latent factor models, which means that the basic neural model is unable to effectively capture the features in news  ... 
arXiv:2110.05792v2 fatcat:wkd45dtj45ge5dodb2cxnggcdu

Probabilistic Machine Learning for Healthcare [article]

Irene Y. Chen, Shalmali Joshi, Marzyeh Ghassemi, Rajesh Ranganath
2020 arXiv   pre-print
Beyond predictive models, we also investigate the utility of probabilistic machine learning models in phenotyping, in generative models for clinical use cases, and in reinforcement learning.  ...  Machine learning can be used to make sense of healthcare data. Probabilistic machine learning models help provide a complete picture of observed data in healthcare.  ...  This work was supported in part by a CIFAR AI Chair at the Vector Institute (MG) and Microsoft Research (MG). LITERATURE CITED  ... 
arXiv:2009.11087v1 fatcat:htosfeqvhndvfmlmud2pvl3nsy
« Previous Showing results 1 — 15 out of 1,517 results