WWN-2: A biologically inspired neural network for concurrent visual attention and recognition

Ji, Zhengping; Weng, Juyang

doi:10.1109/ijcnn.2010.5596778

Cited by 19 publications

(17 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Explaining these developmental effects are out of the scope of this paper, but have been written about elsewhere [22]. Further focusing on learning and development in WWN is [5].…”

Section: B Learning Attentionmentioning

confidence: 99%

“…To train the whole WWN, the following algorithm ran over three iterations per sample. (3) We used a supervised training mechanism ("pulvinar"-based training [5]) to bias V4 to learn foreground patterns: we set its Z based on the firing of the LM area -only neurons with receptive fields on the foreground would receive a top-down boost. (4) For PP (denoted as "P" above) and IT areas, we used 3 × 3 neighborhood updating in the vein of selforganizing maps in order to spread representation throughout the layer.…”

Section: B Learning Attentionmentioning

confidence: 99%

“…Neurons earlier in each pathway represent both location and type in a mixed way. For more information on learning, see [5]. Tests in feedforward mode with a single foreground in the scene showed the recognition and orientation performance improved after epochs of learning (an epoch is an entire round of training all possible samples).…”

Section: B Learning Attentionmentioning

confidence: 99%

“…The second version WWN-2 [5] additionally realized attention and recognition for single objects in natural backgrounds without supplying either position or type (free-viewing), and also used a more complex architecture. Further, all pixel locations were tested.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Where-What Network 3: Developmental top-down attention for multiple foregrounds and complex backgrounds

Luciw

Weng

2010

The 2010 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

Abstract-The Where-What Network (WWN-3) is an artificial developmental network modeled after visual cortical pathways, for the purpose of attention and recognition in the presence of complex natural backgrounds. It is generalpurpose and not pre-determined to detect a certain type of stimulus. It is a learning network, which develops its weights from images using a supervised paradigm and a local Hebbian learning algorithm. Attention has been thought of as bottomup or top-down. This paper focuses on the biologically-inspired mechanisms of top-down attention in WWN-3, through topdown excitation that interacts with bottom-up activity at every layer within the network. Top-down excitation in WWN-3 can control the location of attention by imposing a certain location or disengaging from the current location. It can also control what type of object to search for. Paired layers and sparse coding deal with potential hallucination problems. Top-down attention in WWN occurs as soon as an action emerges at a motor layer, which could be imposed by a teacher or internally selected. Given two competing foregrounds in the same scene, WWN showed effective performance in all the attention modes tested.

show abstract

“…Explaining these developmental effects are out of the scope of this paper, but have been written about elsewhere [22]. Further focusing on learning and development in WWN is [5].…”

Section: B Learning Attentionmentioning

confidence: 99%

Section: B Learning Attentionmentioning

confidence: 99%

Section: B Learning Attentionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Where-What Network 3: Developmental top-down attention for multiple foregrounds and complex backgrounds

Luciw

Weng

2010

The 2010 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

show abstract

“…But only 5 locations were tested. WWN-2 [12] can additionally perform in the mode of freeviewing, realizing the visual attention and object recognition Area Two Area One Area Three without the type or location information and all the pixel locations were tested. WWN-3 [13] can deal with multiple objects in natural backgrounds using arbitrary foreground object contours, not the square contours in WWN-1.…”

Section: Introductionmentioning

confidence: 99%

Skull-closed autonomous development: WWN-6 using natural video

Wang

Weng

2012

The 2012 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

Abstract-While a physical environment interacts with a human individual through the brain's sensors and effectors, internal representations inside the skull-closed brain autonomously emerge and adapt throughout the lifetime. By "skull-closed", we mean that the brain inside the skull is off limit to all teachers in the external physical environment, except the brain's sensory ends and motor ends. We present the Where-What Network 6 (WWN-6), which has realized our goal of fully autonomous development inside a closed network "skull". This means that the human programmer is not allowed to handcraft the internal representation for any fixed extra-body concepts. For example, the meanings of specific values of location or type concept are not known during the programming time. Such meanings emerge through associations imbedded in 'postnatal" experience. This capability is especially challenging when one considers the fact that most elements in the sensory ends are irrelevant to the signals at the effector ends (e.g., many background pixels). How does each vector value in the effector find its corresponding pattern in the correct patch of the sensory image? We outline this autonomous learning theory for the brain and present how the developmental program (DP) of WWN-6 enables the network to perform for attending and recognizing objects in complex backgrounds using natural video. The inputs to the agent (i.e., the network) are not artificially synthesized images as WWNs used before, but drawn from continuous video taken from natural settings where, in general, everything is moving.

show abstract

Where-What Network with CUDA: General Object Recognition and Location in Complex Backgrounds

Wang

Song

et al. 2011

Advances in Neural Networks – ISNN 2011

View full text Add to dashboard Cite

WWN-2: A biologically inspired neural network for concurrent visual attention and recognition

Cited by 19 publications

References 14 publications

Where-What Network 3: Developmental top-down attention for multiple foregrounds and complex backgrounds

Where-What Network 3: Developmental top-down attention for multiple foregrounds and complex backgrounds

Skull-closed autonomous development: WWN-6 using natural video

Where-What Network with CUDA: General Object Recognition and Location in Complex Backgrounds

Contact Info

Product

Resources

About