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..

Camellia: A 128-Bit Block Cipher Suitable for Multiple Platforms - Design and Analysis

Abstract: We present a new 128-bit block cipher called Camellia. Camellia supports 128-bit block size and 128-, 192-, and 256-bit keys, i.e., the same interface specifications as the Advanced Encryption Standard (AES). Efficiency on both software and hardware platforms is a remarkable characteristic of Camellia in addition to its high level of security. It is confirmed that Camellia provides strong security against differential and linear cryptanalyses. Compared to the AES finalists, i.e., MARS, RC6, Rijndael, Serpent, and Twofish, Camellia offers at least comparable encryption speed in software and hardware. An optimized implementation of Camellia in assembly language can encrypt on a Pentium III (800MHz) at the rate of more than 276 Mbits per second, which is much faster than the speed of an optimized DES implementation. In addition, a distinguishing feature is its small hardware design. The hardware design, which includes encryption and decryption and key schedule, occupies approximately 11K gates, which is the smallest among all existing 128-bit block ciphers as far as we know.

New Block Encryption Algorithm MISTY

Abstract: We propose secret-key cryptosystems MISTY1 and MISTY2, which are block ciphers with a 128-bit key, a 64-bit block and a variable number of rounds. MISTY is a generic name for MISTY1 and MISTY2. They are designed on the basis of the theory of provable security against differential and linear cryptanalysis, and moreover they realize high speed encryption on hardware platforms as well as on software environments. Our software implementation shows that MISTY1 with eight rounds can encrypt a data stream in CBC mode at a speed of 20Mbps and 40Mbps on Pentium/100MHz and PA-7200/120MHz, respectively. For its hardware performance, we have produced a prototype LSI by a process of 0.5Μ CMOS gate-array and confirmed a speed of 450Mbps. In this paper, we describe the detailed specifications and design principles of MISTY1 and MISTY2.

Hyper Suprime-Cam

Abstract: Hyper Suprime-Cam (HSC) is an 870 Mega pixel prime focus camera for the 8.2 m Subaru telescope. The wide field corrector delivers sharp image of 0.25 arc-sec FWHM in r-band over the entire 1.5 degree (in diameter) field of view. The collimation of the camera with respect to the optical axis of the primary mirror is realized by hexapod actuators whose mechanical accuracy is few microns. As a result, we expect to have seeing limited image most of the time. Expected median seeing is 0.67 arc-sec FWHM in i-band. The sensor is a p-ch fully depleted CCD of 200 micron thickness (2048 x 4096 15 μm square pixel) and we employ 116 of them to pave the 50 cm focal plane. Minimum interval between exposures is roughly 30 seconds including reading out arrays, transferring data to the control computer and saving them to the hard drive. HSC uniquely features the combination of large primary mirror, wide field of view, sharp image and high sensitivity especially in red. This enables accurate shape measurement of faint galaxies which is critical for planned weak lensing survey to probe the nature of dark energy. The system is being assembled now and will see the first light in August 2012.

A Monte Carlo algorithm for fast projective clustering

Abstract: We propose a mathematical formulation for the notion of optimal projective cluster, starting from natural requirements on the density of points in subspaces. This allows us to develop a Monte Carlo algorithm for iteratively computing projective clusters. We prove that the computed clusters are good with high probability. We implemented a modified version of the algorithm, using heuristics to speed up computation. Our extensive experiments show that our method is significantly more accurate than previous approaches. In particular, we use our techniques to build a classifier for detecting rotated human faces in cluttered images.

CDC: Convolutional-De-Convolutional Networks for Precise Temporal Action Localization in Untrimmed Videos

Abstract: Temporal action localization is an important yet challenging problem. Given a long, untrimmed video consisting of multiple action instances and complex background contents, we need not only to recognize their action categories, but also to localize the start time and end time of each instance. Many state-of-the-art systems use segment-level classifiers to select and rank proposal segments of pre-determined boundaries. However, a desirable model should move beyond segment-level and make dense predictions at a fine granularity in time to determine precise temporal boundaries. To this end, we design a novel Convolutional-De-Convolutional (CDC) network that places CDC filters on top of 3D ConvNets, which have been shown to be effective for abstracting action semantics but reduce the temporal length of the input data. The proposed CDC filter performs the required temporal upsampling and spatial downsampling operations simultaneously to predict actions at the frame-level granularity. It is unique in jointly modeling action semantics in space-time and fine-grained temporal dynamics. We train the CDC network in an end-to-end manner efficiently. Our model not only achieves superior performance in detecting actions in every frame, but also significantly boosts the precision of localizing temporal boundaries. Finally, the CDC network demonstrates a very high efficiency with the ability to process 500 frames per second on a single GPU server. We will update the camera-ready version and publish the source codes online soon.