Consider a time slotted communication channel shared by K active users and a single receiver. It is assumed that the receiver has the ability of the multiple-packet reception (MPR) to correctly receive at most γ (1 ≤ γ < K) simultaneously transmitted packets. Each user accesses the channel following a specific periodical binary sequence, called the protocol sequence, and transmits a packet within a channel slot if and only if the sequence value is equal to one. The fluctuation in throughput is incurred by inevitable random relative shifts among the users due to the lack of feedback. A set of protocol sequences is said to be throughput-invariant (TI) if it can be employed to produce invariant throughput for any relative shifts, i.e., maximize the worst-case throughput. It was shown in the literature that the TI property without considering MPR (i.e., γ = 1) can be achieved by using shift-invariant (SI) sequences, whose generalized Hamming cross-correlation is independent of relative shifts. This paper investigates TI sequences for MPR; results obtained include achievable throughput value, a lower bound on the sequence period, an optimal construction of TI sequences that achieves the lower bound on the sequence period, and intrinsic structure of TI sequences. In addition, we present a practical packet decoding mechanism for TI sequences that incorporates packet header, forward error-correcting code, and advanced physical layer blind signal separation techniques.Collision channel without feedback, protocol sequences, multiple-packet reception, throughput. I. INTRODUCTIONThe idea of using protocol sequences to define deterministic multiaccess protocols for a collision channel without feedback was proposed by Massey and Mathys in [1], and recently has attracted many research revisits [2]-[7] for different design criteria and applications. Compared with time division multiple access (TDMA), ALOHA and carrier sense multiple access (CSMA), protocol sequences do not require stringent synchronization, channel monitoring, backoff algorithm and packet retransmissions. Such a simplicity is particularly desirable in ad hoc networks and sensor networks, in which well-coordinated transmissions and time synchronization may be difficult to achieve due to user mobility, time-varying propagation delays and energy constraints.Moreover, in contrast to the random and contention based schemes, protocol sequences in [2]- [7] can respectively provide a positive short-term throughput guarantee with probability one in the worst case.Previous studies on multiaccess protocols have traditionally assumed a collision channel model of single-packet reception (SPR), in which a packet is received correctly only if it is not involved in a collision, i.e., does not overlap with another. However, the assumption of SPR becomes more and more unsuitable in practice, due to recent advances at reception techniques of the physical (PHY) layer, such as antenna arrays, CDMA technique and beamforming algorithms, which can be employed to ensure that the receive...
Background The electronic health record (EHR) contains a wealth of medical information. An organized EHR can greatly help doctors treat patients. In some cases, only limited patient information is collected to help doctors make treatment decisions. Because EHRs can serve as a reference for this limited information, doctors’ treatment capabilities can be enhanced. Natural language processing and deep learning methods can help organize and translate EHR information into medical knowledge and experience. Objective In this study, we aimed to create a model to extract concept embeddings from EHRs for disease pattern retrieval and further classification tasks. Methods We collected 1,040,989 emergency department visits from the National Taiwan University Hospital Integrated Medical Database and 305,897 samples from the National Hospital and Ambulatory Medical Care Survey Emergency Department data. After data cleansing and preprocessing, the data sets were divided into training, validation, and test sets. We proposed a Transformer-based model to embed EHRs and used Bidirectional Encoder Representations from Transformers (BERT) to extract features from free text and concatenate features with structural data as input to our proposed model. Then, Deep InfoMax (DIM) and Simple Contrastive Learning of Visual Representations (SimCLR) were used for the unsupervised embedding of the disease concept. The pretrained disease concept-embedding model, named EDisease, was further finetuned to adapt to the critical care outcome prediction task. We evaluated the performance of embedding using t-distributed stochastic neighbor embedding (t-SNE) to perform dimension reduction for visualization. The performance of the finetuned predictive model was evaluated against published models using the area under the receiver operating characteristic (AUROC). Results The performance of our model on the outcome prediction had the highest AUROC of 0.876. In the ablation study, the use of a smaller data set or fewer unsupervised methods for pretraining deteriorated the prediction performance. The AUROCs were 0.857, 0.870, and 0.868 for the model without pretraining, the model pretrained by only SimCLR, and the model pretrained by only DIM, respectively. On the smaller finetuning set, the AUROC was 0.815 for the proposed model. Conclusions Through contrastive learning methods, disease concepts can be embedded meaningfully. Moreover, these methods can be used for disease retrieval tasks to enhance clinical practice capabilities. The disease concept model is also suitable as a pretrained model for subsequent prediction tasks.
Let C be a q-ary code of length n and size M, and C(i ) = {c(i) | c = (c(1), c(2), . . . , c(n)) T ∈ C} be the set of ith coordinates of C. The descendant code of a sub-code C ⊆ C is defined to be C (1) × C (2) × · · · × C (n). In this paper, we introduce a multimedia analogue of codes with the identifiable parent property (IPP), called multimedia IPP codes or t-MIPPC(n, M, q), so that given the descendant code of any sub-code C of a multimedia t-IPP code C, one can always identify, as IPP codes do in the generic digital scenario, at least one codeword in C . We first derive a general upper bound on the size M of a multimedia t-IPP code, and then investigate multimedia 3-IPP codes in more detail. We characterize a multimedia 3-IPP code of length 2 in terms of a bipartite graph and a generCommunicated by 123 M. Cheng et al.alized packing, respectively. By means of these combinatorial characterizations, we further derive a tight upper bound on the size of a multimedia 3-IPP code of length 2, and construct several infinite families of (asymptotically) optimal multimedia 3-IPP codes of length 2.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.