IEEE TPAMI 2017
288 papers
The 2017 edition of IEEE Transactions on Pattern Analysis and Machine Intelligence featured 54 papers focusing on 3D reconstruction, manifold learning, and adaptive algorithms. Research themes included hierarchical coarse-to-fine 3D reconstruction, geometric shape descriptors for medical phenotyping, and methods for overcoming topological constraints in mesh reconstruction. Theoretical advancements were made in subspace clustering via vanishing polynomials and solving the pre-image problem in manifold learning.
The volume also emphasized real-time inference and online adaptation, with studies on non-stationary data streams and evolving target representations for visual tracking. Specific technical contributions included the generalization of pooling functions in convolutional neural networks and the development of Kendall and Mallows kernels for permutations. Prolific authors for the year included Qi Tian, Fernando De la Torre, and Jean Ponce. Application-oriented research covered diverse areas such as face recognition through collaborative representation and 3D fingerprint identification using colored LED illumination.
Top Research Directions
Research directions most represented by IEEE TPAMI 2017 papers.
Top Authors
Impactful Papers
Papers from IEEE TPAMI 2017 that shaped research clusters.
Foundational
Kernel Clustering: Density Biases and Solutions
foundational
Uniformizing Data Density to Eliminate G
Learning to Generate Chairs, Tables and Cars with Convolutional Networks
foundational
Overcoming the Annotation Bottleneck in
Influential
Show and Tell: Lessons Learned from the 2015 MSCOCO Image Captioning Challenge.
influential
Beam Search Decoding for Neural Sequence
Learning Supervised Topic Models for Classification and Regression from Crowds
influential
Modeling Correlated Errors in Label Aggr
DASC: Robust Dense Descriptor for Multi-Modal and Multi-Spectral Correspondence Estimation.
influential
Overcoming Affine and Perspective Distor
Bayesian Time-of-Flight for Realtime Shape, Illumination and Albedo.
influential
Resolving Multi-Path Interference in Act
Nuclear Norm Based Matrix Regression with Applications to Face Recognition with Occlusion and Illumi
influential
Mitigating Structural Information Loss f
Cross-Validated Variable Selection in Tree-Based Methods Improves Predictive Performance.
influential
High-Cardinality Bias in Tree-Based Feat
Long-Term Recurrent Convolutional Networks for Visual Recognition and Description
influential
The Vanishing Gradient Problem in Sequen