Get started
Author

Yuma Ichikawa

Statistical-mechanical Study of Deep Boltzmann Machine Given Weight Parameters after Training by Singular Value Decomposition
2022 · Yuma Ichikawa, Koji Hukushima · Journal of the Physical Society of Japan · 8
PDF
Learning Dynamics in Linear VAE: Posterior Collapse Threshold, Superfluous Latent Space Pitfalls, and Speedup with KL Annealing
2023 · Ichikawa, Yuma, Hukushima, Koji · arXiv (Cornell University) · 4
PDF
High-dimensional asymptotics of VAEs: threshold of posterior collapse and dataset-size dependence of rate-distortion curve
2025 · Yuma Ichikawa, Koji Hukushima · Journal of Statistical Mechanics Theory and Experiment · 3
Adaptive Flip Graph Algorithm for Matrix Multiplication
2024 · Yamato Arai, Yuma Ichikawa, et al. · International Symposium on Symbolic and Algebraic Computation · 3
High-dimensional Asymptotics of VAEs: Threshold of Posterior Collapse and Dataset-Size Dependence of Rate-Distortion Curve
2023 · Ichikawa, Yuma, Hukushima, Koji · arXiv (Cornell University) · 2
PDF
Statistical Mechanics of Min-Max Problems
2024 · Ichikawa, Yuma, Hukushima, Koji · arXiv (Cornell University) · 1
PDF
Optimization by Parallel Quasi-Quantum Annealing with Gradient-Based Sampling
2024 · Ichikawa, Yuma, Arai, Yamato · arXiv (Cornell University) · 1
PDF
Controlling Continuous Relaxation for Combinatorial Optimization
2024 · Yuma Ichikawa · neural information processing systems · 1
PDF
Decision Model Factory: Building Specialized Sovereign Decision Models
2026 · Ichikawa, Yuma · arXiv (Cornell University)
PDF
Tensor-network population annealing
2026 · Takumi Oshima, Yuma Ichikawa, et al. · Physical review. E
Kozuchi Agent: A Language-Agnostic Open-Weight Agent for Software Repair
2026 · Mehdi Bahrami, Kosaku Kimura, et al. · IEEE/ACM International Conference on Automated Software Engineering (ASE)
PDF
Beyond Scalar Sensitivity: Activation-Aware Mixed-Precision LLM Quantization with Cross-Layer Refinement
2026 · Yoshida, Akihiro, Ichikawa, Yuma · arXiv (Cornell University)
PDF
LLM-as-an-Improver: Turning Verification into Better Candidates
2026 · Tomihari, Akiyoshi, Ichikawa, Yuma · arXiv (Cornell University)
PDF
From Sweep to Seam: Interleaved Cross-Block Post-Training Quantization
2026 · Jacquemond, Achille, Ichikawa, Yuma, et al. · arXiv (Cornell University)
PDF
One Qubit Can Beat One Bit: Quantum Advantage for Post-Training Quantization
2026 · Ichikawa, Yuma, Mishima, Moeto · arXiv (Cornell University)
PDF
Addressable Recall Compaction for Long Context-Window Control in AI Agents
2026 · Dang, Thang, Ichikawa, Yuma, et al. · arXiv (Cornell University)
PDF
LOGOS: A Living Logic for AI Agent Teams That Evolve With Humans
2026 · Ichikawa, Yuma, Arai, Yamato, et al. · arXiv (Cornell University)
PDF
PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation
2026 · Association for Computational Linguistics 2026, Ichikawa, Yuma, et al. · Underline Science Inc.
Signs Beat Floats: Low-Rank Double-Binary Adaptation for On-Device Fine-Tuning
2026 · Fujisawa, Yoshihiko, Ichikawa, Yuma, et al. · arXiv (Cornell University)
PDF
Data and gnuplot scripts for "Tensor-Network Population Annealing"
2026 · Hukushima, Koji, Ichikawa, Yuma, et al. · Zenodo (CERN European Organization for Nuclear Research)
QQA4CO: Quasi-Quantum Annealing for Combinatorial Optimization
2026 · Ichikawa, Yuma, Arai, Yamato · Zenodo (CERN European Organization for Nuclear Research)
Tensor-Network Population Annealing
2026 · Oshima, Takumi, Ichikawa, Yuma, et al. · arXiv (Cornell University)
PDF
OneComp: One-Line Revolution for Generative AI Model Compression
2026 · Ichikawa, Yuma, Kimura, Keiji, et al. · arXiv (Cornell University)
PDF
Sign Lock-In: Randomly Initialized Weight Signs Persist and Bottleneck Sub-Bit Model Compression
2026 · Sakai, Akira, Ichikawa, Yuma · arXiv (Cornell University)
PDF
PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation
2026 · Yuma Ichikawa, Naoya Takagi, et al. · Proceedings of the conference - Association for Computational Linguistics. Meeting
PDF
Works count: 35
Citation count: 23
H-index: 3
I10-index:
Year
Topic
Type