Mitsubishi Electric

About: Mitsubishi Electric is a company organization based out in Ratingen, Germany. It is known for research contribution in the topics: Signal & Voltage. The organization has 23024 authors who have published 27591 publications receiving 255671 citations. The organization is also known as: Mitsubishi Electric Corporation & Mitsubishi Denki K.K..

A Winnow-Based Approach to Context-Sensitive Spelling Correction

Abstract: A large class of machine-learning problems in natural language require the characterization of linguistic context. Two characteristic properties of such problems are that their feature space is of very high dimensionality, and their target concepts depend on only a small subset of the features in the space. Under such conditions, multiplicative weight-update algorithms such as Winnow have been shown to have exceptionally good theoretical properties. In the work reported here, we present an algorithm combining variants of Winnow and weighted-majority voting, and apply it to a problem in the aforementioned class: context-sensitive spelling correction. This is the task of fixing spelling errors that happen to result in valid words, such as substituting to for too, casual for causal, and so on. We evaluate our algorithm, WinSpell, by comparing it against BaySpell, a statistics-based method representing the state of the art for this task. We find: (1) When run with a full (unpruned) set of features, WinSpell achieves accuracies significantly higher than BaySpell was able to achieve in either the pruned or unpruned conditions (2) When compared with other systems in the literature, WinSpell exhibits the highest performances (3) While several aspects of WinSpell‘s architecture contribute to its superiority over BaySpell, the primary factor is that it is able to learn a better linear separator than BaySpell learnss (4) When run on a test set drawn from a different corpus than the training set was drawn from, WinSpell is better able than BaySpell to adapt, using a strategy we will present that combines supervised learning on the training set with unsupervised learning on the (noisy) test set.

MPEG-4 rate control for multiple video objects

Abstract: This paper describes an algorithm which can achieve a constant bit rate when coding multiple video objects. The implementation is a nontrivial extension of the MPEG-4 rate control algorithm for single video objects which employs a quadratic rate quantizer model. The algorithm is organized into two stages: a pre- and a post-encoding stage. In the pre-encoding stage, an initial target estimate is made for each object. Based on the buffer fullness, the total target is adjusted and then distributed proportional to the relative size, motion, and variance of each object. Based on the new individual targets and rate-quantizer relation for texture, appropriate quantization parameters are calculated. After each object is encoded, the model parameters for each object are updated, and if necessary, frames are skipped to ensure that the buffer does not overflow. A preframeskip control is exercised to avoid buffer overflow when the motion and shape information occupies a significant portion of the bit budget. The rate control algorithm switches between two operation modes so that the coder can reduce the spatial coding accuracy for an improved temporal resolution. A shape-coding control mechanism is also proposed, which provides a tradeoff between texture and shape coding accuracy. Overall, the algorithm is able to successfully achieve the target bit rate, effectively code arbitrarily shaped objects, and maintain a stable buffer level. These techniques have been adopted by the MPEG committee in July 1997 as part of the video verification model (VM8).

On correlation between the order of S-boxes and the strength of DES

Abstract: This paper introduces a practical algorithm for deriving the best differential characteristic and the best linear expression of DES. Its principle is based on a duality between differential cryptanalysis and linear cryptanalysis, and applicable to various block ciphers. Then using this program, we observe how the order of S-boxes affects the strength of DES. We show that the order of the S-boxes is well-arranged against differential cryptanalysis, though it is not the best choice. On the other hand, our experimental results indicate that it is a very weak choice in regard to linear cryptanalysis. In other words, DES can be strengthened by just rearranging the order of the S-boxes.

COLLAGEN: A Collaboration Manager for Software Interface Agents

Abstract: We have implemented an application-independent collaboration manager, called Collagen, based on the SharedPlan theory of discourse, and used it to build a software interface agent for a simple air travel application. The software agent provides intelligent, mixed initiative assistance without requiring natural language understanding. A key benefit of the collaboration manager is the automatic construction of an interaction history which is hierarchically structured according to the user‘s and agent‘s goals and intentions.

Advances in joint CTC-attention based end-to-end speech recognition with a deep CNN encoder and RNN-LM

Abstract: We present a state-of-the-art end-to-end Automatic Speech Recognition (ASR) model. We learn to listen and write characters with a joint Connectionist Temporal Classification (CTC) and attention-based encoder-decoder network. The encoder is a deep Convolutional Neural Network (CNN) based on the VGG network. The CTC network sits on top of the encoder and is jointly trained with the attention-based decoder. During the beam search process, we combine the CTC predictions, the attention-based decoder predictions and a separately trained LSTM language model. We achieve a 5-10\% error reduction compared to prior systems on spontaneous Japanese and Chinese speech, and our end-to-end model beats out traditional hybrid ASR systems.