An information-theoretic lower bound is developed for the caching system studied by Maddah-Ali and Niesen. By comparing proposed with decentralized coded scheme of Niesen, optimal memory-rate tradeoff characterized to within a multiplicative gap 4.7 worst case, improving previous analytical 12. Furthermore, case when users' requests follow uniform distribution, tightened 4.7, 72. As an independent result interest, single-user average in which user multiple files, it proved that most...

Under the paradigm of caching, partial data is delivered before actual requests users are known. In this paper, problem modeled as a canonical distributed source coding with side information, where information represents users' requests. For single-user case, single-letter characterization optimal rate region established, and for several important special cases, closed-form solutions given, including scenario uniformly user it shown that caching strategy closely related to total correlation...

Improved lower bounds are derived on the average and worst case rate-memory tradeoffs of Maddah-Ali Niesen-coded caching scenario. For any number users files for arbitrary cache sizes, multiplicative gap between exact tradeoff new bound is shown to be less than 2.315 in scenario 2.507 average-case

As face recognition is widely used in diverse security-critical applications, the study of anti-spoofing (FAS) has attracted more and attention. Several FAS methods have achieved promising performance if attack types testing data are included training data, while significantly degrades for unseen types. It essential to learn generalized discriminative features prevent overfitting pre-defined spoof This paper proposes a novel dual-stage disentangled representation learning method that can...

In this paper, we consider a cache aided network in which each user is assumed to have individual caches, while upon users' requests, an update message sent through common link all users. First, formulate general information theoretic setting that represents the database as discrete memoryless source, and requests side available everywhere except at encoder. The decoders' objective recover function of source information. By viewing networks terms distributed coding problem arguments, present...

Function computation over Gaussian networks with orthogonal components is studied for arbitrarily correlated discrete memoryless sources. Two classes of functions are considered: 1) the arithmetic sum function and 2) type function. The in this paper defined as a set multiple weighted sums, which includes averaging sources estimating each special cases. or frequency histogram counts number occurrences argument, yields various fundamental statistics, such mean, variance, maximum, minimum,...

An information-theoretic lower bound is developed for the caching system studied by Maddah-Ali and Niesen. By comparing proposed with decentralized coded scheme of Niesen, optimal memory--rate tradeoff characterized to within a multiplicative gap $4.7$ worst case, improving previous analytical $12$. Furthermore, case when users' requests follow uniform distribution, tightened $4.7$, $72$. As an independent result interest, single-user average in which user multiple files, it proved that most...

This paper investigates the downlink of a cloud radio access network (C-RAN) in which central processor communicates with two mobile users through base stations (BSs). The BSs act as relay nodes and cooperate each other error-free rate-limited links. We develop analyze coding schemes for this scenario. first scheme modifies Liu-Kang (to make it amenable to rigorous analysis) extends introduce common codewords apply C-RAN BS-to-BS cooperation. enables arbitrary correlation among auxiliary...

In this paper, we present information theoretic inner and outer bounds on the fundamental tradeoff between cache memory size update rate in a multi-user network. Each user is assumed to have individual caches, while upon users' requests, an message sent though common link all users. The database represented as discrete memoryless source request side that available at decoders encoder, but oblivious encoder. We establish two bounds, first based centralized caching strategy second...

Improved lower bounds on the worst-case and average-case rate-memory tradeoffs for Maddah-Ali&Niesen coded-caching scenario are presented. For any number of users files arbitrary cache sizes, multiplicative gap between exact tradeoff new bound is less than 2.315 in 2.507 scenario.

The fading AWGN two-user two-hop network is considered where the channel coefficients are independent and identically distributed (i.i.d.) according to a continuous distribution vary over time. For broad class of distributions, ergodic sum capacity characterized within constant number bits/second/hertz, signal-to-noise ratio. achievability follows from analysis an interference neutralization scheme relays partitioned into M pairs, neutralized separately by each pair relays. When = 1,...

Although significant progress has been made in face recognition, demographic bias still exists recognition systems. For instance, it usually happens that the performance for a certain group is lower than others. In this paper, we propose MixFairFace framework to improve fairness models. First of all, argue commonly used attribute-based metric not appropriate recognition. A system can only be considered fair while every person close performance. Hence, new evaluation protocol fairly evaluate...

Weakly-Supervised Semantic Segmentation (WSSS) aims to train segmentation models using image data with only image-level supervision. Since precise pixel-level annotations are not accessible, existing methods typically focus on producing pseudo masks for training by refining CAM-like heatmaps. However, the produced heatmaps may capture discriminative regions of object categories or associated co-occurring backgrounds. To address issues, we propose a Prompt Learning WSSS (SemPLeS) framework,...

Motivated by the caching problem introduced Maddah-Ali and Niesen, a of distributed source coding with side information is formulated, which captures distinct interesting aspect caching. For single-user case, single-letter characterization optimal rate region presented. cases where composed either independent or nested components, exact regions are found some intuitive strategies confirmed to be optimal. When components arbitrarily correlated uniform requests, strategy closely related total...

In wireless sensor networks, various applications involve learning one or multiple functions of the measurements observed by sensors, rather than themselves. This paper focuses on computation type-threshold which include maximum, minimum, and indicator as special cases. Previous work studied this problem under collocated collision network model showed that many probabilistic models for measurements, achievable rates tend to zero number sensors increases. paper, networks are modeled fully...

We consider a fading AWGN 2-user 2-hop network in which the channel coefficients are independently and identically distributed (i.i.d.) drawn from continuous distribution vary over time. For broad class of distributions, we characterize ergodic sum capacity within constant number bits/sec/Hz, independent signal-to-noise ratio. The achievability follows analysis an interference neutralization scheme where relays partitioned into K pairs, is neutralized separately by each pair relays. = 1,...

This work investigates the downlink of cloud radio access networks (C-RANs), assuming digital cooperation links among base stations (BSs). A generalization data-sharing scheme is proposed for case two BSs and mobile users. The generalized includes a common part allows full exploitation correlation auxiliary codewords. between are used to exchange redirect indices precomputed at central processor. On other hand, by simplifying achievable rate region distributed decode-forward (DDF) scheme, it...

In wireless sensor networks, various applications involve learning one or multiple functions of the measurements observed by sensors, rather than themselves. This paper focuses on class type-threshold functions, e.g., maximum and indicator functions. A simple network model capturing both broadcast superposition properties channels is considered: collocated Gaussian network. general multiround coding scheme exploiting interaction (through broadcast) developed. Through careful scheduling...

This paper presents a log-domain decoder for non-binary LDPC over GF(q). Comparing with the conventional O(q <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) decoders, proposed can efficiently reduce decoding complexity to O(qlogq) only negligible degradation in BER. Comparisons on both simulated BER performance and computational between existing decoders are also provided.

Deep learning approaches have achieved highly accurate face recognition by training the models with very large image datasets. Unlike availability of 2D datasets, there is a lack 3D datasets available to public. Existing public were usually collected few subjects, leading over-fitting problem. This paper proposes two CNN improve RGB-D task. The first segmentation-aware depth estimation network, called DepthNet, which estimates maps from RGB images including semantic segmentation information...

Holistic driving scene understanding is a critical step toward intelligent transportation systems. It involves different levels of analysis, interpretation, reasoning and decision making. In this paper, we propose 3D dynamic analysis framework as the first understanding. Specifically, given sequence synchronized 2D sensory data, systematically integrates perception modules to obtain position, orientation, velocity category traffic participants ego car in reconstructed semantically labeled...

The problem of successive refinement in distributed source coding and joint source-channel is considered. emphasis placed on the case where sources have to be recovered losslessly second stage. In coding, it shown that all are successively refinable sum rate, with respect any (joint) distortion measure first assumed independent only a (per letter) function For class multiple access channels, linear functions. Finally, when equal entropy, simple sufficient condition refinability provided for...