A vision architecture for unconstrained and incremental learning of multiple categories
Citations
53 citations
Cites methods from "A vision architecture for unconstra..."
...Additionally Θmin is the node-dependent learning rate as proposed by Kirstein et al. (2008):...
[...]
...Finally the long-term stability of these incrementally learned representation nodes is considered as proposed by Kirstein et al. (2008). Additionally for our learning approach a category-specific forward feature selection method is used to enable the separation of co-occurring categories, because it defines category-specific metrical “views” on the nodes of the exemplar-based network....
[...]
...Furthermore we recently could show that our proposed cLVQ learning method can be integrated into a larger vision system that allows online learning of categories based on hand-held and complex-shaped objects under full rotation (Kirstein et al., 2008, 2009)....
[...]
41 citations
Cites background from "A vision architecture for unconstra..."
...Different systems focus on different aspects of the problem, such as the system architecture and integration [68], [69], [71], learning [66], [67], [71], or social interaction [70]....
[...]
41 citations
Cites background from "A vision architecture for unconstra..."
...Different systems focus on different aspects of this problem, such as the system architecture and integration [3], [4], [6], learning [1], [2], [6], [7], or social interaction [5]....
[...]
35 citations
Cites background from "A vision architecture for unconstra..."
...Because of the intuitive definition of models in terms of prototypical representatives, prototype-based methods like LVQ enjoy a wide popularity in application domains, particularly if human inspection and interaction are necessary, or life-long model adaptation is considered [28, 20, 18]....
[...]
20 citations
Cites background from "A vision architecture for unconstra..."
...Different systems focus on different aspects of this problem, such as the system architecture and integration (Bauckhage et al., 2001; Billard & Hayes, 1999; Briggs & Scheutz, 2012; Bolder et al., 2008; Hawes et al., 2010; Karaoguz, Rodemann, Wrede, & Goerick, 2012; Kirstein et al., 2009; Lallee et al., 2012; Lutkebohle et al., 2009; Mason & Lopes, 2011; Sun, 2007); learning and symbol grounding (Salvi, Montesano, Bernardino, & Santos-Victor, 2012; Roy & Pentland, 2002; Billard & Hayes, 1999; Steels & Kaplan, 2000; Kirstein et al....
[...]
...…and symbol grounding (Salvi, Montesano, Bernardino, & Santos-Victor, 2012; Roy & Pentland, 2002; Billard & Hayes, 1999; Steels & Kaplan, 2000; Kirstein et al., 2009; de Greeff, Delaunay, & Belpaeme, 2009; Chernova & Veloso, 2009; Belpaeme & Morse, 2012; Briggs & Scheutz, 2012; Tellex,…...
[...]
..., 2009; Mason & Lopes, 2011; Sun, 2007); learning and symbol grounding (Salvi, Montesano, Bernardino, & Santos-Victor, 2012; Roy & Pentland, 2002; Billard & Hayes, 1999; Steels & Kaplan, 2000; Kirstein et al., 2009; de Greeff, Delaunay, & Belpaeme, 2009; Chernova & Veloso, 2009; Belpaeme & Morse, 2012; Briggs & Scheutz, 2012; Tellex, Thaker, Joseph, & Roy, 2014; Perera & Allen, 2013; Schiebener, Morimoto, Asfour, & Ude, 2013; Deits et al., 2013); motivation (Lutkebohle et al....
[...]
...…et al., 2001; Billard & Hayes, 1999; Briggs & Scheutz, 2012; Bolder et al., 2008; Hawes et al., 2010; Karaoguz, Rodemann, Wrede, & Goerick, 2012; Kirstein et al., 2009; Lallee et al., 2012; Lutkebohle et al., 2009; Mason & Lopes, 2011; Sun, 2007); learning and symbol grounding (Salvi,…...
[...]
References
46,906 citations
14,509 citations
"A vision architecture for unconstra..." refers methods in this paper
...For this learning method we propose a combination of an incremental exemplar-based network and a forward feature selection method (see (Guyon & Elissee 2003) for an introduction to feature selection methods)....
[...]
8,197 citations
"A vision architecture for unconstra..." refers methods in this paper
...(10) Each wkmin(c)(rl) is updated based on the standard LVQ learning rule (Kohonen 1989), but is restricted to feature dimensions f ∈ Sc: w kmin(c) f := w kmin(c) f + µ Θ kmin(c)(rlf − w kmin(c) f ) ∀f ∈ Sc, (11) where µ = 1 if the categorization decision for rl was correct, otherwise µ = −1 and…...
[...]
...For our learning system a novel incremental category learning method is proposed that combines a learning vector quantization (LVQ) (Kohonen 1989) network to approach the “stability-plasticity dilemma” with a category-specific forward feature selection....
[...]
[...]
5,672 citations
4,713 citations
"A vision architecture for unconstra..." refers methods in this paper
...We use a feed-forward feature extraction architecture inspired by the Neocognitron (Fukushima 1980) to extract shape features....
[...]
Related Papers (5)
Frequently Asked Questions (16)
Q2. What is the purpose of the forward feature selection method?
The forward feature selection method is used to find low dimensional subsets of category-specific features by predominately selecting features, which occur almost exclusively for a certain category.
Q3. Why do the authors remove the noise from the initial foreground hypothesis?
Due to the fact that the objects are presented by hand, skin color parts in the segment are systematic noise, which the authors remove from the initial foreground hypothesis based on the detection method proposed by Fritsch et al. (2002).
Q4. What is the main drawback of those architectures commonly used for identification tasks?
The major drawback of those architectures commonly used for identification tasks is the inefficient separation of cooccuring categories.
Q5. What is the reason for the indepen-dent representation of categories?
It seems to be that for their categorization task the indepen-dent representation of categories somehow weakens the forgetting effect of SLP networks.
Q6. What is the common strategy for life-long learning architectures?
A common strategy for life-long learning architectures e.g. (Hamker, 2001; Furao & Hasegawa, 2006; Kirstein et al., 2008) is the usage of a node specific learning rate combined with an incremental node insertion rule.
Q7. What are the advantages of these approaches?
The advantages of these approaches are their robustness against partial occlusion, scale changes, and the ability to deal with cluttered environments.
Q8. What is the reason for the poor performance of color categories?
For color categories the effect of imprecise foreground masks on the categorization performance seems also to be only minor, otherwise the performance would be considerably lower.
Q9. What is the way to test the representation of objects?
This allows that object views can be first used to test the STM and LTM representation and after providing confirmed labels the same views can also be used to enhance the representation by transferring them into the STM, even if they where recorded before the confirmation.
Q10. How is the learning of the category specific LTM based on the current available feature vectors?
Based on the currently available feature vectors, the learning methods are used to incorporate this STM knowledge into the LTM by applying the learning dynamics of the cLVQ method described in Section 4.2.3.
Q11. How do the authors relax the separation of object views?
To relax this separation and to make the most efficient use of object views, the authors introduce a sensory memory concept for temporarily remembering views of the currently attended object, by using the same one-shot learning method as used for the STM.
Q12. What is the computation of local edge responses?
(1)This computation of local edge responses is restricted to the positions in the foreground mask with ξi(x, y) > 0, whereas the ∗ denotes the inner product of two vectors.
Q13. What is the main problem when dealing with learning in unconstrained environments?
One of the essential problems when dealing with learning in unconstrained environments is the definition of a shared attention concept between the learning system and the human tutor.
Q14. How is the hypothesis list communicated to the user?
Additionally the hypothesis list is repeatedly communicated to the user (in 5 second intervals), while newly acquired segments are also used to refine this list.
Q15. What is the effect of the constraint on the learning task?
Additionally this constraint strongly reduces the appearance variations of the presented objects and therefore makes the category learning task much easier.
Q16. How does the learning system perform the necessary processing steps?
The authors could show that their learning system can efficiently perform all necessary processing steps including figure-ground segregation, feature extraction and incremental learning.