Fels SS, Hinton GE (1990), Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study. In Diaper D, Gilmore D, Cockton G, Shackel B (Editors) Human-Computer Interaction: Interact, Proceedings of the IFIP Third International Conference on Human-Computer Interaction, North-Holland, Oxford: 683-687.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....needs to be very carefully thought out for tasks where user s motions are intended to carry out a meaning. In this latter case, hardwiring virtual world actions to specific sensor values forces commitments that would risk reducing device expressiveness and can make applications difficult to use [10]. Adaptive pattern recognition can be used to overcome these problems, by letting the definition of the mapping between sensor measurements and actions in the virtual world be more complex, and therefore increasing the expressive power of the devices. Furthermore, the possibility of specifying ....

Fels SS, Hinton GE (1990), Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study. In Diaper D, Gilmore D, Cockton G, Shackel B (Editors) Human-Computer Interaction: Interact, Proceedings of the IFIP Third International Conference on Human-Computer Interaction, North-Holland, Oxford: 683-687.

....(see Future Work) As the move to dynamic gestures is considered, it is the Author s opinion that neural networks would provide a more suitable solution. It has already been shown that American Sign Language (ASL) can successfully be recognised through the use of a data glove and a neural network (Fels, 1990). Working with a PDM provides a significantly smaller dimensional problem than that of a data glove with its multiple flexation sensors and therefore again demonstrates the beauty simplicity of the PDM. Ahmad et al. proposed a similar approach to recognising gestures in a model of a hand placed ....

S. S. Fels and G. E. Hinton (1990), Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study, Human-Computer Interaction - INTERACT `90, Diaper, D et al (eds), pp 683-688, Elsevier Science Publishers B.V. (North-Holland).

....it, but needs to be very carefully thought out for tasks where user s motions are intended to carry out a meaning. In this latter case, hardwiring virtual world actions to specific sensor values forces commitments that risk reducing device expressiveness and can make applications difficult to use (Fels and Hinton 1990). In order to overcome these problems, mediator objects can be interposed between sensors and models to transform the information accordingly to interaction metaphors. Two major types of mediators are used in VB2: adaptive pattern recognizers, to enhance sensor data with classification ....

Fels SS, Hinton GE (1990) Building Adaptive Interfaces with Neural Networks: The GloveTalk Pilot Study. In Diaper D, Gilmore D, Cockton G, Shackel B (Editors) HumanComputer Interaction: Interact, Proceedings of the IFIP Third International Conference on Human-Computer Interaction, North-Holland, Oxford, pp.683-687.

....it, but needs to be very carefully thought out for tasks where user s motions are intended to carry out a meaning. In this latter case, hardwiring virtual world actions to specific sensor values forces commitments that risk reducing device expressiveness and can make applications difficult to use (Fels and Hinton, 1990). In order to overcome these problems, mediator objects can be interposed between sensors and models to transform the information accordingly to interaction metaphors. Two major types of mediators are used in VB2: adaptive pattern recognizers, to enhance sensor data with classification ....

Fels SS, Hinton GE (1990) Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study. In Diaper D, Gilmore D, Cockton G, Shackel B (Editors) Human-Computer Interaction: Interact, Proceedings of the IFIP Third International Conference on Human-Computer Interaction, North-Holland, Oxford: 683-687.

....has been mentioned was to use a SL recognition system to remotely control a multijointed artificial arm Published research in this area has thus concentrated on finger spelling devices, which are tedious to use for all except proper nouns. Grimes [14] and Kramer [23] developed such systems. Fels [10] went one step further, mapping postures to a vocabulary of root words with the direction and velocity of hand motion providing word modifiers, such as endings and stresses. Tamura 3 6 dimensions of wrist pose, 4 joints for each finger and thumb, 2 joints in the wrist, 2 at the elbow and 3 in ....

....is no formal basis for constructing neural networks and the topology, unit activation functions, learning strategy and learning rate all must be determined by trial and error. Despite these drawbacks, somewhat successful gesture recognition has been performed with neural networks, notably by Fels [10]. 2.3.1 Fels GloveTalk His work concentrated on building a gesture to speech interface, using a VPL DataGlove connected to a DECtalk speech synthesiser via a series of neural networks. The neural networks that adapt the mappings for a user from training examples of gestures must be small to ....

S. Sidney Fels and Geoffrey E. Hinton. Building adaptive interfaces with neural networks: The glove-talk pilot study. In Human-Computer Interaction---INTERACT ' 90, pages 683--688. IFIP, Elsevier Science Publishers B.V. (North-Holland), 1990.

....needs to be very carefully thought out for tasks where user s motions are intended to carry out a meaning. In this latter case, hardwiring virtual world actions to specific sensor values forces commitments that would risk reducing device expressiveness and can make applications difficult to use [10]. Adaptive pattern recognition can be used to overcome these problems, by letting the definition of the mapping between sensor measurements and actions in the virtual world be more complex, and therefore increasing the expressive power of the devices. Furthermore, the possibility of specifying ....

Fels SS, Hinton GE (1990), Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study. In Diaper D, Gilmore D, Cockton G, Shackel B (Editors) Human-Computer Interaction: Interact, Proceedings of the IFIP Third International Conference on Human-Computer Interaction, North-Holland, Oxford: 683-687.

....neither in space nor time which they call postures and between dynamic hand and finger movements which they call gestures. Zimmerman [Zimmerman 1987] also notes the difference between static gestures (a peace sign) and dynamic gestures (waving goodbye) There seems to be common agreement [Fels 1990, Hauptmann 1989, Murakami 1991, Pausch 1991, Weber 1989, Weimer 1989] that, in the case of hand gestures, the term gesture should be applied to a continuous dynamic hand movement of both the fingers and hand position in space. 2.2 Why use Gestures as Input There already exist many other types ....

....It is noteworthy that several projects concerned with gesture recognition have used sign language as a basis to develop gestural input [Beale 1990, Kramer 1988, Murakami 1991 and Takahashi 1991] 2. Review 6 Philip A. Harling Gesture Input using Neural Networks In their Glove Talk study, Fels and Hinton [Fels 1990] stress the importance of having the input device closely match the requirements of the task being performed, i.e. having the right tool for the job. For example, this would suggest that a three dimensional space tracker is better for indicating points in space than a keyboard would be. Work by ....

[Article contains additional citation context not shown here]

Fels 1990 S. Sidney Fels and Geoffrey E. Hinton, Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study, in Human-Computer Interaction: Interact `90, pp. 683-688 (1990).

....kinds of interfaces based on 3 dimensional gesticulative input. When a hand tracking device is incorporated into a computer environment, the users gain the freedom to explore their virtual surroundings using gesticulative input. There are at least three approaches to use this freedom: KL87] and [Fel90] used sign language approaches; WGW90] and [PEF 90] used object based manipulations; Bol80] and [FMHR86] used voice assisted methods. Because we believe Direct Manipulation [Shn92] is a good paradigm for data manipulation, we have focused on the second approach. However, the VPA still ....

S. Sidney Fels. Building Adaptive Interfaces with Neural Networks: the Glove-Talk Pilot Study. Technical Report CRG-TR-90-1, Dept. Computer Science, U. Toronto, February 1990.

....for a root word, the normal ending is used. The 66 root words and six endings were extracted from the 850 word vocabulary of Basic English [4] The following sections describe the five neural networks used. Full details of the complete system, the training data, and the performance can be found in [5]. Learning in a multilayer neural network All five neural networks were of the feedforward variety with one hidden layer. They were trained using backpropagation [6] In a feedforward network, each unit has an activity level that is determined by the input received from units in the layer below. ....

....increases the number of misses. The results for the training and test performance of the root network are summarized in table 3. More details of how the training data were collected and the effects of changing the architecture of the network or the input and output encodings are described in [5]. INSERT FIGURE 5 ABOUT HERE) INSERT TABLE 3 ABOUT HERE) Once the strobe network is working it is simple to create training and test data for the other networks. Targets (combinations of the word root, ending, rate, and stress) are presented to the user and s he simply makes the appropriate ....

S.S. Fels, Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study, Technical Report CRG-TR-90-1, University of Toronto, Toronto, Canada, 1990.

No context found.

Fels, S.; & Hinton, G. "Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study." Human-Computer Interaction: Interact `90. Amsterdam: North-Holland, 1990, pp. 683-688.

No context found.

Fels,S.S., Hinton,G.E., "Building Adaptive Interfaces with Neural Networks: The Glove--Talk Pilot Study," INTERACT'90 Proceedings, pp. 683--688, Elsevier, 1990.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC