Institution
Qualcomm
Company•Farnborough, United Kingdom•
About: Qualcomm is a company organization based out in Farnborough, United Kingdom. It is known for research contribution in the topics: Wireless & Signal. The organization has 19408 authors who have published 38405 publications receiving 804693 citations. The organization is also known as: Qualcomm Incorporated & Qualcomm, Inc..
Papers published on a yearly basis
Papers
More filters
•
25 Jan 2007
TL;DR: In this paper, a wireless communication device can automatically adapt to environmental conditions such as light, sound, motion, temperature and the like and, at user discretion, adjust functionality on the device that is affected by these conditions.
Abstract: Methods, devices, apparatus and systems are provided for adaptation of a wireless communication device based on current environmental conditions, historical environmental condition patterns and/or historical usage patterns. In this regard, the device senses such environmental conditions as light, sound, motion, temperature and the like and, automatically or at user discretion, adjusts functionality on the device that is affected by these conditions. Additionally, by determining environmental and usage patterns the device can adapt to the patterns by providing certain functionality and/or applications to the user at prescribed times of the day, week, month, etc. By being able to automatically adapt the functionality of a wireless communication device, the user of such a device experiences greater ease in using the device and an overall heightened user experience.
136 citations
•
TL;DR: AdaRound is proposed, a better weight-rounding mechanism for post-training quantization that adapts to the data and the task loss that outperforms rounding-to-nearest by a significant margin and establishes a new state-of-the-art forPost- training quantization on several networks and tasks.
Abstract: When quantizing neural networks, assigning each floating-point weight to its nearest fixed-point value is the predominant approach. We find that, perhaps surprisingly, this is not the best we can do. In this paper, we propose AdaRound, a better weight-rounding mechanism for post-training quantization that adapts to the data and the task loss. AdaRound is fast, does not require fine-tuning of the network, and only uses a small amount of unlabelled data. We start by theoretically analyzing the rounding problem for a pre-trained neural network. By approximating the task loss with a Taylor series expansion, the rounding task is posed as a quadratic unconstrained binary optimization problem. We simplify this to a layer-wise local loss and propose to optimize this loss with a soft relaxation. AdaRound not only outperforms rounding-to-nearest by a significant margin but also establishes a new state-of-the-art for post-training quantization on several networks and tasks. Without fine-tuning, we can quantize the weights of Resnet18 and Resnet50 to 4 bits while staying within an accuracy loss of 1%.
136 citations
•
07 Mar 2012TL;DR: In this article, the authors proposed a harmonization of the scan order to code both the significance map of the transform coefficients as well as to code the levels of a transform coefficient.
Abstract: This disclosure describes techniques for coding transform coefficients associated with a block of residual video data in a video coding process. Aspects of this disclosure include the selection of a scan order for both significance map coding and level coding, as well as the selection of contexts for entropy coding consistent with the selected scan order. This disclosure proposes a harmonization of the scan order to code both the significance map of the transform coefficients as well as to code the levels of the transform coefficient. It is proposed that the scan order for the significance map should be in the inverse direction (i.e., from the higher frequencies to the lower frequencies). This disclosure also proposes that transform coefficients be scanned in subsets as opposed to fixed subblocks. In particular, transform coefficients are scanned in a subset consisting of a number of consecutive coefficients according to the scan order.
136 citations
•
26 Sep 1997TL;DR: In this article, a system and method for determining the position of a user terminal (for example, a mobile wireless telephone) in a low-Earth orbit satellite communications system is presented.
Abstract: A system and method for determining the position of a user terminal (for example, a mobile wireless telephone) in a low-Earth orbit satellite communications system (100). The system includes a user terminal (106), at least one satellite (104) with a known position and velocity, and a gateway (102) (that is, a terrestrial base station) for communicating with the user terminal (106) through the satellite (104). The method includes the steps of determining a range parameter (802) and a range-rate parameter (806). A range parameter represents a distance between the satellite and the user terminal. A range-rate parameter represents a relative radial velocity between that satellite and the user terminal. The position of the user terminal on the Earth's surface is then determined (810) based on the range parameter, the range-rate parameter, and the known position and velocity of the satellite.
136 citations
•
13 Dec 2004TL;DR: In this paper, an access point in a multi-antenna system broadcasts data using spatial spreading to randomize an effective channel observed by each user terminal for each block of data symbols broadcast by the access point.
Abstract: An access point in a multi-antenna system broadcasts data using spatial spreading to randomize an 'effective' channel observed by each user terminal for each block of data symbols broadcast by the access point. At the access point, data is coded, interleaved, and modulated to obtain ND data symbol blocks to be broadcast in NM transmission spans, where ND ≥1 and NM >1. The ND data symbol blocks are partitioned into NM data symbol subblocks, one subblock for each transmission span. A steering matrix is selected (e.g., in a deterministic or pseudo-random manner from among a set of L steering matrices) for each subblock. Each data symbol subblock is spatially processed with the steering matrix selected for that subblock to obtain transmit symbols, which are further processed and broadcast via NT transmit antennas and in one transmission span to user terminals within a broadcast coverage area.
136 citations
Authors
Showing all 19413 results
Name | H-index | Papers | Citations |
---|---|---|---|
Jian Yang | 142 | 1818 | 111166 |
Xiaodong Wang | 135 | 1573 | 117552 |
Jeffrey G. Andrews | 110 | 562 | 63334 |
Martin Vetterli | 105 | 761 | 57825 |
Vinod Menon | 101 | 269 | 60241 |
Michael I. Miller | 92 | 599 | 34915 |
David Tse | 92 | 438 | 67248 |
Kannan Ramchandran | 91 | 592 | 34845 |
Michael Luby | 89 | 282 | 34894 |
Max Welling | 89 | 441 | 64602 |
R. Srikant | 84 | 432 | 26439 |
Jiaya Jia | 80 | 294 | 33545 |
Hai Li | 79 | 570 | 33848 |
Simon Haykin | 77 | 454 | 62085 |
Christopher W. Bielawski | 76 | 334 | 32512 |