Reverse TDNN: An Architecture For Trajectory Generation
نویسندگان
چکیده
Yann Le Cun AT&T Bell Laboratories 101 Crawford Corner Rd Holmdel, NJ 07733 The backpropagation algorithm can be used for both recognition and generation of time trajectories. When used as a recognizer, it has been shown that the performance of a network can be greatly improved by adding structure to the architecture. The same is true in trajectory generation. In particular a new architecture corresponding to a "reversed" TDNN is proposed. Results show dramatic improvement of performance in the generation of hand-written characters. A combination of TDNN and reversed TDNN for compact encoding is also suggested.
منابع مشابه
Stacked Long-Term TDNN for Spoken Language Recognition
This paper introduces a stacked architecture that uses a time delay neural network (TDNN) to model long-term patterns for spoken language identification. The first component of the architecture is a feed-forward neural network with a bottleneck layer that is trained to classify context-dependent phone states (senones). The second component is a TDNN that takes the output of the bottleneck, conc...
متن کاملOptimal Trajectory Generation for Energy Consumption Minimization and Moving Obstacle Avoidance of SURENA III Robot’s Arm
In this paper, trajectory generation for the 4 DOF arm of SURENA III humanoid robot with the purpose of optimizing energy and avoiding a moving obstacle is presented. For this purpose, first, kinematic equations for a seven DOF manipulator are derived. Then, using the Lagrange method, an explicit dynamics model for the arm is developed. In the next step, in order to generate the desired traject...
متن کاملMulti-State Time Delay Neural Networks for Continuous Speech Recognition
Alex Waibel Carnegie Mellon University Pittsburgh, PA 15213 [email protected] We present the "Multi-State Time Delay Neural Network" (MS-TDNN) as an extension of the TDNN to robust word recognition. Unlike most other hybrid methods. the MS-TDNN embeds an alignment search procedure into the connectionist architecture. and allows for word level supervision. The resulting system has the ability to ma...
متن کاملReverse Electrodialysis for Salinity Gradient Power Generation: Challenges and Future Perspectives
Salinity gradient energy, which is also known as Blue energy, is a renewable energy form that can be extracted from the mixing of two solutions with different salinities. About 80% of the current global electricity demand could potentially be covered by this energy source. Among several energy extraction technologie...
متن کاملOptimal Trajectory Generation for a Robotic Worm via Parameterization by B-Spline Curves
In this paper we intend to generate some set of optimal trajectories according to the number of control points has been applied for parameterizing those using B-spline curves. The trajectories are used to generate an optimal locomotion gait in a crawling worm-like robot. Due to gait design considerations it is desired to minimize the required torques in a cycle of gait. Similar to caterpillars,...
متن کامل