About

Kenneth Rose is a Distinguished Professor in the Department of Electrical and Computer Engineering at UC Santa Barbara. His research interests include Information Theory, Source Coding and Networking, Distributed Coding, 360-degree Video Coding, 3D Audio Coding, Pattern Recognition and Machine Learning, and Nonconvex Optimization. He is associated with the Signal Compression Lab and is involved in advancing knowledge and technology in these areas. Professor Rose's work focuses on developing innovative solutions in data compression, multimedia coding, and machine learning, contributing to the fields of information theory and signal processing.

Research topics

• Computer Science
• Artificial Intelligence
• Medicine
• Computer vision
• Transport engineering
• Psychology
• Algorithm
• Environmental health
• Telecommunications
• Biology
• Mathematics
• Engineering
• Cell biology

Selected publications

• Adaptive Upscaling Filter Selection with Wiener Filter Compression Techniques in Reference Picture Resampling

  2025-05-25

  articleSenior author

  Encoding video frames at adaptive resolutions has been shown to be an effective strategy, as it balances the trade-off between side information and residual data based on frame size and content. Reference Picture Resampling (RPR) in the Versatile Video Coding (H.266/VVC) standard allows for motion compensation between frames with different resolutions, overcoming the limitation that resolution switching points must occur at intra random-access points (IRAP), and that all frames within a closed GOP must be encoded at a single resolution. Frame rescaling is a critical step in RPR; however, applying a single upscaling filter to all video frames of different resolutions is not ideal. In this paper, an adaptive upscaling filter selection scheme is proposed, which incorporates Wiener filter trained at the frame level. The rate-distortion (RD) performance of various upscaling filters is compared to determine the optimal adaptive filter. The Wiener filter coefficients are reduced and compressed using predictive coding. Experimental results on VVC test model (VTM) 23.3 show average BD rate gains of 0.33% for 4K, 2.15% for 1080p and 2.11% for 480p video sequences at a 2.0x RPR scaling ratio under random-access configurations. At a 1.5x RPR scaling ratio, the BD rate gains are 0.57% for 4K, 2.22% for 1080p and 4.71% for 480p video sequences.

  PublisherDOI

• On Ultra Low-Delay Compression of Higher Order Ambisonics Signals

  2024-03-19

  articleSenior author

  The challenge of coding delay is becoming increasingly recognized as a barrier to the broader implementation of Higher Order Ambisonics (HOA), which is a highly flexible format for spatial audio encoding and reproduction. A low latency is essential for various applications, such as virtual reality, interactive gaming, and live online music sessions. Unfortunately, significant strides in the development of spatial audio codecs often compromise on latency to enhance compression efficiency. This work presents a low-delay codec designed for the compression of HOA signals. The codec uses a combination of singular value decomposition, short-term linear prediction, cascaded long-term prediction, and sub-band coding, as well as entropy coding, in order to maximally compress HOA signals while maintaining a low delay. A variety of configurations allow for algorithmic delays between 54 samples to 206 samples, at the cost of bitrate given a fixed quality level. The proposed codec outperforms the low-delay implementation of a standard codec for HOA lossy compression, both in terms of delay and bitrate at medium and higher quality levels.

  PublisherDOI

• Ultra-Low Delay Lossless Compression of Higher Order Ambisonics

  2024-03-18 · 2 citations

  articleSenior author

  Coding delay has emerged as a significant obstacle to wider adoption of Higher Order Ambisonics (HOA), a promising flexible format employed in representation and coding of spatial audio. Low delay is critical in many applications, such as virtual reality, gaming, and online music performance, a good subset of which further require lossless compression. However, major advances in spatial audio codec design often sacrifice low delay in order to maximize compression gains. We present a new lossless, ultra-low delay HOA codec, optimized to take advantage of both the spatial and temporal redundancies in HOA data, while maintaining an extremely low delay. The presented codec has an algorithmic delay of one single sample, at the cost of increased computational complexity. This codec outperforms the state-of-the-art for lossless, low-delay HOA coding, both in terms of delay and bitrate.

  PublisherDOI

• StegaNeRV: Video Steganography using Implicit Neural Representation

  2024-06-17 · 8 citations

  article

  Numerous studies have recently advanced the state-of-the art for representing videos through an implicit neural network (INR). As these models become increasingly ubiquitous, there is a growing demand for concealing data within INR reconstructed videos such as for storing content metadata and sensitive licensing information. In this paper, we explore a new space in video steganography, hiding a distinct image within each RGB frame output by an INR. We propose a joint training strategy of a U-Net based steganographic decoder with an INR model for video. Experimental results show that hidden images can be embedded and subsequently reconstructed with high fidelity while preserving the quality of the cover frames. Furthermore we demonstrate that by introducing an attention module which emphasizes hiding within the edges and rich texture patches in the cover frame, secret images can be reconstructed with superior quality and can also be concealed at greater resolutions.

  PublisherDOI

• Alternate Learning and Compression Approaching R(D)

  arXiv (Cornell University) · 2024-11-05

  preprintOpen accessSenior author

  The inherent trade-off in on-line learning is between exploration and exploitation. A good balance between these two (conflicting) goals can achieve a better long-term performance. Can we define an optimal balance? We propose to study this question through a backward-adaptive lossy compression system, which exhibits a "natural" trade-off between exploration and exploitation.

  PublisherOA PDFDOI

• Emergency Department Visits for Pedestrians Injured in Motor Vehicle Traffic Crashes — United States, January 2021–December 2023

  MMWR Morbidity and Mortality Weekly Report · 2024-05-02 · 2 citations

  articleOpen access

  Traffic-related pedestrian deaths in the United States reached a 40-year high in 2021. Each year, pedestrians also suffer nonfatal traffic-related injuries requiring medical treatment. Near real-time emergency department visit data from CDC's National Syndromic Surveillance Program during January 2021-December 2023 indicated that among approximately 301 million visits identified, 137,325 involved a pedestrian injury (overall visit proportion = 45.62 per 100,000 visits). The proportions of visits for pedestrian injury were 1.53-2.47 times as high among six racial and ethnic minority groups as that among non-Hispanic White persons. Compared with persons aged ≥65 years, proportions among those aged 15-24 and 25-34 years were 2.83 and 2.61 times as high, respectively. The visit proportion was 1.93 times as high among males as among females, and 1.21 times as high during September-November as during June-August. Timely pedestrian injury data can help collaborating federal, state, and local partners rapidly monitor trends, identify disparities, and implement strategies supporting the Safe System approach, a framework for preventing traffic injuries among all road users.

  PublisherOA PDFDOI

• Low Dose Carbon Monoxide Is Neuroprotective in Models of Parkinson’s Disease (S42.008)

  Neurology · 2023-04-25 · 3 citations

  article

  <h3>Objective:</h3> To evaluate the neuroprotective potential of low dose CO treatment in Parkinson disease (PD) models. <h3>Background:</h3> Despite the many risks of smoking tobacco, the risk of PD is markedly reduced among smokers and may derive from biological activity of specific tobacco smoke constituent(s). On the basis of accumulating evidence supporting the therapeutic potential of CO in other disease contexts, this study set out to determine whether low dose CO is neuroprotective in PD models. <h3>Design/Methods:</h3> In an AAV-alpha-synuclein (aSyn) model, rats underwent right nigral injection of AAV1/2-asynA53T and left injection of empty AAV, followed by treatment with oral CO drug product (HBI-002 10ml/kg, daily by gavage) or vehicle. In a short-term MPTP model (40mg/kg, i.p.), mice were treated with inhaled CO (iCO) (250ppm) or air. In an in vitro rotenone model, SH-SY5Y cells exposed to rotenone were treated with CO (200 ppm) or air. All analyses, including immunohistochemistry for nigral aSyn and tyrosine hydroxylase, HPLC measurement of striatal dopamine, stereological cell counting, and biochemical analyses were conducted blinded to treatment condition. <h3>Results:</h3> Each HBI-002 treatment increased carboxy-hemoglobin to 6%. In the aSyn model, treatment with HBI-002 reduced ipsilateral loss of both striatal dopamine and TH-positive neurons in the substantia nigra pars compacta, and HBI-002 reduced aSyn aggregates and aSyn S129 phosphorylation. In the MPTP model, treatment with low dose iCO also reduced loss of striatal dopamine and loss of TH+ neurons. In saline-treated mice, iCO had no effect on striatal dopamine or TH+ cell counts. HBI-002 upregulated heme oxygenase-1, HIF-1a, and cathepsin D. In the rotenone model, treatment with low dose CO reduced cell death. <h3>Conclusions:</h3> These results demonstrating reduced cell death and aSyn pathology in animal and cell models support the therapeutic potential of low dose CO for PD. <b>Disclosure:</b> Dr. Gomperts has received personal compensation in the range of $0-$499 for serving as a Consultant for EIP. Dr. Gomperts has received personal compensation in the range of $0-$499 for serving on a Scientific Advisory or Data Safety Monitoring board for Jannsen. The institution of Dr. Gomperts has received research support from NIH. The institution of Dr. Gomperts has received research support from LBDA. The institution of Dr. Gomperts has received research support from DOD/CDMRP. The institution of Dr. Gomperts has received research support from FFFPRI. The institution of Dr. Gomperts has received research support from NIH. The institution of Dr. Gomperts has received research support from MJFF. The institution of Dr. Gomperts has received research support from NIH. The institution of Dr. Gomperts has received research support from NIH. The institution of Dr. Gomperts has received research support from MJFF. The institution of Dr. Gomperts has received research support from NIH. The institution of Dr. Gomperts has received research support from NIH. Dr. Rose has nothing to disclose. Dr. Zorlu has nothing to disclose. Dr. Xue has nothing to disclose. Dr. Cai has received personal compensation for serving as an employee of Pfizer . Ms. Lin has nothing to disclose. Mr. Lee has nothing to disclose. Dr. Gomperts has stock in Hillhurst Biopharmaceuticals inc. The institution of Dr. Gomperts has received research support from Hillhurst Biopharmaceuticals inc. Dr. Gomperts has received intellectual property interests from a discovery or technology relating to health care. The institution of Dr. Schwarzschild has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Bial Biotech (indirectly, as a service of the Parkinson Study Group service). The institution of Dr. Schwarzschild has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Biogen (indirectly, as a service of the Parkinson Study Group service). The institution of Dr. Schwarzschild has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for UCB (indirectly, as a service of the Parkinson Study Group service). Dr. Schwarzschild has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Eli Lilly. The institution of Dr. Schwarzschild has received research support from NIH. The institution of Dr. Schwarzschild has received research support from Parkinson’s Foundation. The institution of Dr. Schwarzschild has received research support from Michael J Fox Foundation. The institution of Dr. Schwarzschild has received research support from Farmer Family Foundation. Dr. Schwarzschild has a non-compensated relationship as a Chair, Executive Committee with the Parkinson Study Group that is relevant to AAN interests or activities. Dr. Chen has nothing to disclose.

  PublisherDOI

• Comment on egusphere-2022-1248

  2023-03-31

  peer-reviewOpen access1st authorCorresponding

  <strong class="journal-contentHeaderColor">Abstract.</strong> Stocks of Atlantic cod, <em>Gadus Morhua</em>, show diverse recovery responses when fishing pressure is relieved. The expected outcome of reduced fishing pressure is that the population regains its size. However, there are also cod stocks that seem to be locked in a state of low abundance from which population growth does not, or only slowly, occur. A plausible explanation for this phenomenon can be provided by the Allee effect, which takes place when recruitment per capita is positively related to population density or abundance. However, because of methodological limitations and data constraints, such a phenomenon is often perceived as being rare or non-existent in marine fish. In this study, we used time-series of 17 Atlantic cod stocks to fit a family of population equations that consider the abundance of spawners, their body weight as well as sea water temperature as independent components of recruitment. The developed stock-recruitment function disentangles the effects of spawner abundance, spawner weight and temperature on recruitment dynamics and captures the diversity of density dependencies (compensation, Allee effect) of the recruitment production in Atlantic cod. The results show for 13 cod stocks an inherent, spawner abundance related Allee effect. Allee effect strength, i.e. the relative change between maximum and minimum recruitment per capita at low abundance, was <em>increased</em> when recruitment production was suppressed by unfavorable changes in water temperature and/or in spawner weight. The latter can be a concomitant of heavy fishing or a result of temperature related altered body growth. Allee effect strength was <em>decreased</em> when spawner weight and/or temperature elevated recruitment production. We show how anthropogenic stress can increase the risk of Allee effects in stocks where ocean temperature and/or spawner weight had been beneficial in the past, but are likely to &ldquo;unmask&rdquo; and strengthen an inherent Allee effect under future conditions.

  PublisherOA PDFDOI

• Transform Domain Temporal Prediction for Dynamic Point Cloud Compression

  2023-09-27

  articleSenior author

  With the advent of immersive multimedia technologies, interest in dynamic point cloud compression standards, such as video-based point cloud compression (V-PCC), has been steadily growing. At the core of V-PCC, conventional video codecs such as High Efficiency Video Coding are employed for encoding the geometry video, attribute video and the occupancy map. Standard video codecs perform temporal prediction using motion compensation by simple pixel-copying from the reference frame. However, conventional pixel-domain prediction does not exploit spatial correlation within the block. This is highly sub-optimal especially due to large homogeneous regions in the geometry video. In this paper, spatial decorrelation is first achieved via the Discrete Cosine Transform. Prediction is then performed in the transform domain, where the observed variation in temporal correlation across frequencies is exploited for a more accurate prediction. Design of the correlation filters required for this procedure poses challenges. The inherent instability in closed-loop design is tackled using the Asymptotic Closed Loop paradigm which uses an iterative open-loop algorithm that asymptotically converges to closed-loop operation. Experimental results on geometry videos generated from common test conditions show significant bitrate savings of up to 5.8% (about 3% on average) and highlight the efficacy of our approach.

  PublisherDOI

• A Stochastic Rate-Distortion Approach to Supervised Learning Systems

  2023-06-25

  articleSenior author

  Machine learning applications have exploded in recent years due to the availability of huge data sets as well as advances in computational and storage capabilities. Although successful methods have been proposed to reduce learning system complexity while maintaining required accuracy levels, theoretical understanding of the underlying trade-offs remains elusive. In this paper, the classical supervised learning problem is reformulated within a rate-distortion framework. It provides insights into crucial accuracy-complexity trade-offs, by considering the overall learning system as consisting of two components. The first is tasked with extracting (learning) from the source the minimal number of information bits necessary to ultimately achieve the prescribed output accuracy. The learned bits are then used to retrieve the desired output from the second component, an appropriately designed codebook. The premise here is that an optimal system is characterized by having to learn the minimum amount of information from the source, just sufficient to yield the system output at the desired precision, which implies efficiency in terms of system complexity, generalization and training data requirements. The design and training of such a reformulated system is detailed in this paper, and asymptotically optimal performance achieving the rate-distortion bound is established.

  PublisherDOI

Recent grants

• Fast Approximate Search and Retrieval of High-Dimensional Data

  NSF · $390k · 2004–2008

• CIF: Small: An Integrated Framework for Distributed Source Coding and Dispersive Information Routing

  NSF · $419k · 2010–2014

• CIF: Small: Analog Networking: Distributed Source-Channel Approaches to Delay and Resource Constrained Communications

  NSF · $302k · 2011–2015

• CIF: Small: A Resource-Scalable Unifying Framework for Aural Signal Coding

  NSF · $498k · 2009–2013

• CIF: Small: The Common Information Framework and Optimal Coding for Layered Storage and Transmission of Audio Signals

  NSF · $495k · 2013–2016

Frequent coauthors

• Emrah Akyol

  Binghamton University

  82 shared

• Kumar Viswanatha

  University of California, Santa Barbara

  46 shared

• Tejaswi Nanjundaswamy

  Mayachitra (United States)

  45 shared

• S.L. Regunathan

  32 shared

• Vinay Melkote

  30 shared

• Ertem Tuncel

  University of California, Riverside

  27 shared

• David J. Miller

  26 shared

• Zurab Khasidashvili

  Intel (Israel)

  25 shared

Labs

• Signal Compression LabPI