arXiv subjects

Yasuhiko Nakashima

Publications and source records attributed to Yasuhiko Nakashima.

29 records · Page 2Linked to original sources

arXiv2025

Lifelong Whole Slide Image Analysis: Online Vision-Language Adaptation and Past-to-Present Gradient Distillation

Whole Slide Images (WSIs) play a crucial role in accurate cancer diagnosis and prognosis, as they provide tissue details at the cellular level. However, the rapid growth of computational tasks involving WSIs poses significant challenges. Given that WSIs are gigapixels in size, they present difficulties in terms of storage, processing, and model training. Therefore, it is essential to develop lifelong learning approaches for WSI analysis. In scenarios where slides are distributed across multiple institutes, we aim to leverage them to develop a unified online model as a computational tool for cancer diagnosis in clinical and hospital settings. In this study, we introduce ADaFGrad, a method designed to enhance lifelong learning for whole-slide image (WSI) analysis. First, we leverage pathology vision-language foundation models to develop a framework that enables interaction between a slide's regional tissue features and a predefined text-based prototype buffer. Additionally, we propose a gradient-distillation mechanism that mimics the gradient of a logit with respect to the classification-head parameters across past and current iterations in a continual-learning setting. We construct a sequence of six TCGA datasets for training and evaluation. Experimental results show that ADaFGrad outperforms both state-of-the-art WSI-specific and conventional continual-learning methods after only a few training epochs, exceeding them by up to +5.068% in the class-incremental learning scenario while exhibiting the least forgetting (i.e., retaining the most knowledge from previous tasks). Moreover, ADaFGrad surpasses its baseline by as much as +40.084% in accuracy, further demonstrating the effectiveness of the proposed modules.

cs.CV↗

arXiv2025

ZeroSlide: Is Zero-Shot Classification Adequate for Lifelong Learning in Whole-Slide Image Analysis in the Era of Pathology Vision-Language Foundation Models?

Lifelong learning for whole slide images (WSIs) poses the challenge of training a unified model to perform multiple WSI-related tasks, such as cancer subtyping and tumor classification, in a distributed, continual fashion. This is a practical and applicable problem in clinics and hospitals, as WSIs are large, require storage, processing, and transfer time. Training new models whenever new tasks are defined is time-consuming. Recent work has applied regularization- and rehearsal-based methods to this setting. However, the rise of vision-language foundation models that align diagnostic text with pathology images raises the question: are these models alone sufficient for lifelong WSI learning using zero-shot classification, or is further investigation into continual learning strategies needed to improve performance? To our knowledge, this is the first study to compare conventional continual-learning approaches with vision-language zero-shot classification for WSIs. Our source code and experimental results will be available soon.

cs.CV↗

arXiv2025

FQsun: A Configurable Wave Function-Based Quantum Emulator for Power-Efficient Quantum Simulations

Quantum computers are promising powerful computers for solving complex problems, but access to real quantum hardware remains limited due to high costs. Although the software simulators on CPUs/GPUs such as Qiskit, ProjectQ, and Qsun offer flexibility and support for many qubits, they struggle with high power consumption and limited processing speed, especially as qubit counts scale. Accordingly, quantum emulators implemented on dedicated hardware, such as FPGAs and analog circuits, offer a promising path for addressing energy efficiency concerns. However, existing studies on hardware-based emulators still face challenges in terms of limited flexibility and lack of fidelity evaluation. To overcome these gaps, we propose FQsun, a quantum emulator that enhances performance by integrating four key innovations: efficient memory organization, a configurable Quantum Gate Unit (QGU), optimized scheduling, and multiple number precisions. Five FQsun versions with different number precisions are implemented on the Xilinx ZCU102, consuming a maximum power of 2.41W. Experimental results demonstrate high fidelity, low mean square error, and high normalized gate speed, particularly with 32-bit versions, establishing FQsun's capability as a precise quantum emulator. Benchmarking on famous quantum algorithms reveals that FQsun achieves a superior power-delay product, outperforming software simulators on CPUs in the processing speed range.

quant-ph↗

arXiv2025

Parallelizing the stabilizer formalism for quantum machine learning applications

The quantum machine learning model is emerging as a new model that merges quantum computing and machine learning. Simulating very deep quantum machine learning models requires a lot of resources, increasing exponentially based on the number of qubits and polynomially based on the depth value. Almost all related works use state-vector-based simulators due to their parallelization and scalability. Extended stabilizer formalism simulators solve the same problem with fewer computations because they act on stabilizers rather than long vectors. However, the gate application sequential property leads to less popularity and poor performance. In this work, we parallelize the process, making it feasible to deploy on multi-core devices. The results show that the proposal implementation on Python is faster than Qiskit, the current fastest simulator, 4.23 times in the case of 4-qubits, 60,2K gates.

quant-ph↗

arXiv2024

Exploring the Limitations of Kolmogorov-Arnold Networks in Classification: Insights to Software Training and Hardware Implementation

Kolmogorov-Arnold Networks (KANs), a novel type of neural network, have recently gained popularity and attention due to the ability to substitute multi-layer perceptions (MLPs) in artificial intelligence (AI) with higher accuracy and interoperability. However, KAN assessment is still limited and cannot provide an in-depth analysis of a specific domain. Furthermore, no study has been conducted on the implementation of KANs in hardware design, which would directly demonstrate whether KANs are truly superior to MLPs in practical applications. As a result, in this paper, we focus on verifying KANs for classification issues, which are a common but significant topic in AI using four different types of datasets. Furthermore, the corresponding hardware implementation is considered using the Vitis high-level synthesis (HLS) tool. To the best of our knowledge, this is the first article to implement hardware for KAN. The results indicate that KANs cannot achieve more accuracy than MLPs in high complex datasets while utilizing substantially higher hardware resources. Therefore, MLP remains an effective approach for achieving accuracy and efficiency in software and hardware implementation.

cs.LG↗

arXiv2019

A Compact Low-Latency Systematic Successive Cancellation Polar Decoder for Visible Light Communication Systems

Channel polarization and Polar code are widely considered as major breakthroughs in coding theory because they have shown promising features for future wireless standards. The main drawbacks of Polar code are high-latency in decoding hardware, and unimpressive error-correction performance in case limited code-length is implemented. These two disadvantages limit implementation of Polar code in low-throughput wireless communication systems. In this paper, we propose a low-complexity low-latency hardware architecture for the soft-decision compact (16,11) Systematic Successive Cancellation Polar Decoder (S-SCD). Experimental results has shown that the latency of the proposed S-SCD improves 3.75 times and 2.75 times compared with conventional and 2b-SC architectures. Besides, it has also shown a better BER/FER performance compared with RS(15,11) code, which is applied widely in current VLC-based systems.

eess.SP↗

arXiv2019

Non-RLL DC-Balance based on a Pre-scrambled Polar Encoder for Beacon-based Visible Light Communication Systems

Current flicker mitigation (or DC-balance) solutions based on run-length limited (RLL) decoding algorithms are high in complexity, suffer from reduced code rates, or are limited in application to hard-decoding forward error correction (FEC) decoders. Fortunately, non-RLL DC-balance solutions can overcome the drawbacks of RLL-based algorithms, but they meet some difficulties in system latency, low code rate or inferior error-correction performance. Recently, non-RLL flicker mitigation solution based on Polar code has proved to be a most optimal approach due to its natural equal probabilities of short runs of 1's and 0's with high error-correction performance. However, we found that this solution can only maintain DC balance only when the data frame length is sufficiently long. Therefore, these solutions are not suitable for using in beacon-based visible light communication (VLC) systems, which usually transmit ID information in small-size data frames. In this paper, we introduce a flicker mitigation solution designed for beacon-based VLC systems that combines a simple pre-scrambler with a (256;158) non-systematic polar encoder.

cs.IT↗

arXiv2018

Hardware Implementation of A Non-RLL Soft-decoding Beacon-based Visible Light Communication Receiver

Visible light communication (VLC)-based beacon systems, which usually transmit identification (ID) information in small-size data frames are applied widely in indoor localization applications. There is one fact that flicker of LED light should be avoid in any VLC systems. Current flicker mitigation solutions based on run-length limited (RLL) codes suffer from reduced code rates, or are limited to hard-decoding forward error correction (FEC) decoders. Recently, soft-decoding techniques of RLL-codes are proposed to support soft-decoding FEC algorithms, but they contain potentials of high-complexity and time-consuming computations. Fortunately, non-RLL direct current (DC)-balance solutions can overcome the drawbacks of RLL-based algorithms, however, they meet some difficulties in system latency or inferior error-correction performances. Recently, non-RLL flicker mitigation solution based on Polar code has proved to be an optimal approach due to its natural equal probabilities of short runs of 1's and 0's with high error-correction performance. However, we found that this solution can only maintain the DC balance only when the data frame length is sufficiently long. Accordingly, short beacon-based data frames might still be a big challenge for flicker mitigation in such non-RLL cases. In this paper, we introduce a flicker mitigation solution designed for VLC-based beacon systems that combines a simple pre-scrambler with a Polar encoder which has a codeword smaller than the previous work 8 times. We also propose a hardware architecture for the proposed compact non-RLL VLC receiver for the first time. Also, a 3-bit soft-decision filter is introduce to enable soft-decoding of Polar decoder to improve the performance of the receiver.

eess.SP↗

arXiv2018

VLSI Architecture of Compact Non-RLL Beacon-based Visible Light Communication Transmitter and Receiver

In this paper, we introduce a couple of hardware implementations of compact VLC transmitter and receiver for the first time. Compared with related works, our VLC transmitter is non-RLL one, that means flicker mitigation can be guaranteed even without RLL codes. In particular, we have utilized a centralized bit probability distribution of a prescrambler and a Polar encoder to create a non-RLL flicker mitigation solution. Moreover, at the receiver, a 3-bit soft-decision filter is proposed to analyze signals received from real VLC channel to extract log-likelihood ratio (LLR) values and feed them to the FEC decoder. Therefore, soft-decoding of Polar decoder can be implemented to improve the bit-error-rate (BER) performance of the VLC system. Finally, we introduce a novel very large scale integration (VLSI) architecture for the compact VLC transmitter and receiver; and synthesis our design under FPGA/ASIC synthesis tools. Due to the non-RLL basic, our system has an evidently good code-rate and a reduced-complexity compared with other RLL-based receiver works. Also, we present FPGA and ASIC synthesis results of the proposed architecture with evaluations of power consumption, area, energy-per-bits and so on.

eess.SP↗

arXiv2012

Quantum Walks on the Line with Phase Parameters

In this paper, a study on discrete-time coined quantum walks on the line is presented. Clear mathematical foundations are still lacking for this quantum walk model. As a step towards this objective, the following question is being addressed: {\it Given a graph, what is the probability that a quantum walk arrives at a given vertex after some number of steps?} This is a very natural question, and for random walks it can be answered by several different combinatorial arguments. For quantum walks this is a highly non-trivial task. Furthermore, this was only achieved before for one specific coin operator (Hadamard operator) for walks on the line. Even considering only walks on lines, generalizing these computations to a general SU(2) coin operator is a complex task. The main contribution is a closed-form formula for the amplitudes of the state of the walk (which includes the question above) for a general symmetric SU(2) operator for walks on the line. To this end, a coin operator with parameters that alters the phase of the state of the walk is defined. Then, closed-form solutions are computed by means of Fourier analysis and asymptotic approximation methods. We also present some basic properties of the walk which can be deducted using weak convergence theorems for quantum walks. In particular, the support of the induced probability distribution of the walk is calculated. Then, it is shown how changing the parameters in the coin operator affects the resulting probability distribution.

quant-ph↗

arXiv2012

Tensor Rank and Strong Quantum Nondeterminism in Multiparty Communication

In this paper we study quantum nondeterminism in multiparty communication. There are three (possibly) different types of nondeterminism in quantum computation: i) strong, ii) weak with classical proofs, and iii) weak with quantum proofs. Here we focus on the first one. A strong quantum nondeterministic protocol accepts a correct input with positive probability, and rejects an incorrect input with probability 1. In this work we relate strong quantum nondeterministic multiparty communication complexity to the rank of the communication tensor in the Number-On-Forehead and Number-In-Hand models. In particular, by extending the definition proposed by de Wolf to {\it nondeterministic tensor-rank} ($nrank$), we show that for any boolean function $f$ when there is no prior shared entanglement between the players, 1) in the Number-On-Forehead model, the cost is upper-bounded by the logarithm of $nrank(f)$; 2) in the Number-In-Hand model, the cost is lower-bounded by the logarithm of $nrank(f)$. Furthermore, we show that when the number of players is $o(\log\log n)$ we have that $NQP\nsubseteq BQP$ for Number-On-Forehead communication.

cs.CC↗