2019

Unbiased Evaluation of Deep Metric Learning Algorithms

Fehervari, Istvan, Ravichandran, Avinash, Appalaraju, Srikar

Understand

Deep metric learning (DML) is a popular approach for images retrieval, solving verification (same or not) problems and addressing open set classification.

Arguably, the most common DML approach is with triplet loss, despite significant advances in the area of DML.
Triplet loss suffers from several issues such as collapse of the embeddings, high sensitivity to sampling schemes and more importantly a lack of performance when compared to more modern methods.
We attribute this adoption to a lack of fair comparisons between various methods and the difficulty in adopting them for novel problem statements.

Built on

Bagging predictors
L. Breiman · 1996
Earlier work this paper cites.
Random forests
L. Breiman · 2001
Earlier work this paper cites.
Greedy function approximation: a gradient boosting machine
J. H. Friedman · 2001
Earlier work this paper cites.
Learning a similarity metric discriminatively, with application to face verification
S. Chopra, R. Hadsell, and Y. LeCun · 2005
Earlier work this paper cites.
The Caltech-UCSD Birds-200-2011 Dataset
C. Wah, S. Branson, P. Welinder, P. Perona, and S. Belongie · 2011
Earlier work this paper cites.
Google image swirl: A large-scale content-based image visualization system
Y. Jing, H. A. Rowley, J. Wang, D. Tsai, C. Rosenberg, and M. Covell · 2012
Earlier work this paper cites.
Learning mixtures of submodular shells with application to document summarization
H. Lin and J. Bilmes · 2012
Earlier work this paper cites.
Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude
T. Tieleman and G. Hinton · 2012
Earlier work this paper cites.
3d object representations for fine-grained categorization
J. Krause, M. Stark, J. Deng, and L. Fei-Fei · 2013
Earlier work this paper cites.
Mxnet: A flexible and efficient machine learning library for heterogeneous distributed systems
Original
T. Chen, M. Li, Y. Li, M. Lin, N. Wang, M. Wang, T. Xiao, B. Xu, C. Zhang, and Z. Zhang · 2015
Earlier work this paper cites.
Deep metric learning using triplet network
E. Hoffer and N. Ailon · 2015
Earlier work this paper cites.
Batch normalization: Accelerating deep network training by reducing internal covariate shift
S. Ioffe and C. Szegedy · 2015
Earlier work this paper cites.

Similar

Visual search at pinterest
Y. Jing, D. Liu, D. Kislyuk, A. Zhai, J. Xu, J. Donahue, and S. Tavel · 2015
Cited alongside, same era.
Facenet: A unified embedding for face recognition and clustering
F. Schroff, D. Kalenichenko, and J. Philbin · 2015
Cited alongside, same era.
Deep residual learning for image recognition
K. He, X. Zhang, S. Ren, and J. Sun · 2016
Cited alongside, same era.
Deep metric learning via lifted structured feature embedding
H. Oh Song, Y. Xiang, S. Jegelka, and S. Savarese · 2016
Cited alongside, same era.
Metric learning with adaptive density discrimination
O. Rippel, M. Paluri, P. Dollár, and L. D. Bourdev · 2016
Cited alongside, same era.
Improved deep metric learning with multi-class n-pair loss objective
K. Sohn · 2016
Cited alongside, same era.
Rethinking the inception architecture for computer vision
C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna · 2016
Cited alongside, same era.
Deep spectral clustering learning
M. T. Law, R. Urtasun, and R. S. Zemel · 2017
Cited alongside, same era.
No fuss distance metric learning using proxies
Y. Movshovitz-Attias, A. Toshev, T. K. Leung, S. Ioffe, and S. Singh · 2017
Cited alongside, same era.
Deep metric learning via facility location
H. Oh Song, S. Jegelka, V. Rathod, and K. Murphy · 2017
Cited alongside, same era.
Bier-boosting independent embeddings robustly
M. Opitz, G. Waltner, H. Possegger, and H. Bischof · 2017
Cited alongside, same era.
Prototypical networks for few-shot learning
J. Snell, K. Swersky, and R. Zemel · 2017
Cited alongside, same era.

Then

Deep metric learning with angular loss
J. Wang, F. Zhou, S. Wen, X. Liu, and Y. Lin · 2017
Later among the works it cites.
Sampling matters in deep embedding learning
C.-Y. Wu, R. Manmatha, A. J. Smola, and P. Krahenbuhl · 2017
Later among the works it cites.
Hard-aware deeply cascaded embedding
Y. Yuan, K. Yang, and C. Zhang · 2017
Later among the works it cites.
Deep metric learning with hierarchical triplet loss
W. Ge, W. Huang, D. Dong, and M. R. Scott · 2018
Later among the works it cites.
Attention-based ensemble for deep metric learning
W. Kim, B. Goyal, K. Chawla, J. Lee, and K. Kwon · 2018
Later among the works it cites.
Collaborative deep metric learning for video understanding
J. Lee, S. Abu-El-Haija, B. Varadarajan, and A. P. Natsev · 2018
Later among the works it cites.
Deep metric learning with bier: Boosting independent embeddings robustly
M. Opitz, G. Waltner, H. Possegger, and H. Bischof · 2018
Later among the works it cites.
Deep randomized ensembles for metric learning
H. Xuan, R. Souvenir, and R. Pless · 2018
Later among the works it cites.
Making classification competitive for deep metric learning
Original
A. Zhai and H. Wu · 2018
Later among the works it cites.
Scalable logo recognition using proxies
I. Fehérvári and S. Appalaraju · 2019
Closest in time.
Few-shot learning with embedded class models and shot-free meta training
Original
A. Ravichandran, R. Bhotika, and S. Soatto · 2019
Closest in time.
Ranked list loss for deep metric learning
X. Wang, Y. Hua, E. Kodirov, G. Hu, R. Garnier, and N. M. Robertson · 2019
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…