arXiv subjects

Hongyi Huang

Publications and source records attributed to Hongyi Huang.

10 recordsLinked to original sources

arXiv2026

RetailAgent: Structured Adverse Timing in Self-Conditioned Multimodal LLM Trading Agents

In financial markets, a sequential policy that reacts systematically to price movements may become predictable to other market participants. This paper studies whether large language model (LLM) agents exhibit such directional structure through RetailAgent, an experimental framework in which an LLM observes anonymized intraday equity price histories and permitted state, then repeatedly chooses long (hold the stock) or flat (stay out) before the subsequent interval return is revealed. We compare returns during long and flat intervals along the same stock's intraday path after removing the overall fraction of long decisions. This exposure-matched measure reveals persistent negative timing across modality, horizon, state, and model family. Shuffling saved action sequences substantially attenuates the effect, showing that alignment between actions and subsequent returns drives the negative score. Feeding self-authored memories into decisions further increases policy persistence, while timing becomes more negative among stock-days on which the agent uses both actions. These results reveal stable, recoverable directional structure in sequential LLM financial decisions and a behavioral signal for studying how another participant could respond to a predictable policy.

cs.AI↗

arXiv2026

CS-MUNet: A Channel-Spatial Dual-Stream Mamba Network for Multi-Organ Segmentation

Recently Mamba-based methods have shown promise in abdominal organ segmentation. However, existing approaches neglect cross-channel anatomical semantic collaboration and lack explicit boundary-aware feature fusion mechanisms. To address these limitations, we propose CS-MUNet with two purpose-built modules. The Boundary-Aware State Mamba module employs a Bayesian-attention framework to generate pixel-level boundary posterior maps, injected directly into Mamba's core scan parameters to embed boundary awareness into the SSM state transition mechanism, while dual-branch weight allocation enables complementary modulation between global and local structural representations. The Channel Mamba State Aggregation module redefines the channel dimension as the SSM sequence dimension to explicitly model cross-channel anatomical semantic collaboration in a data-driven manner. Experiments on two public benchmarks demonstrate that CS-MUNet consistently outperforms state-of-the-art methods across multiple metrics, establishing a new SSM modeling paradigm that jointly addresses channel semantic collaboration and boundary-aware feature fusion for abdominal multi-organ segmentation.

cs.CV↗

arXiv2025

Towards Analyzing and Understanding the Limitations of VAPO: A Theoretical Perspective

The VAPO framework has demonstrated significant empirical success in enhancing the efficiency and reliability of reinforcement learning for long chain-of-thought (CoT) reasoning tasks with large language models (LLMs). By systematically addressing challenges such as value model bias, heterogeneous sequence lengths, and sparse reward signals, VAPO achieves state-of-the-art performance. While its practical benefits are evident, a deeper theoretical understanding of its underlying mechanisms and potential limitations is crucial for guiding future advancements. This paper aims to initiate such a discussion by exploring VAPO from a theoretical perspective, highlighting areas where its assumptions might be challenged and where further investigation could yield more robust and generalizable reasoning agents. We delve into the intricacies of value function approximation in complex reasoning spaces, the optimality of adaptive advantage estimation, the impact of token-level optimization, and the enduring challenges of exploration and generalization.

cs.LG↗

arXiv2025

VQ-Logits: Compressing the Output Bottleneck of Large Language Models via Vector Quantized Logits

Large Language Models (LLMs) have achieved remarkable success but face significant computational and memory challenges, particularly due to their extensive output vocabularies. The final linear projection layer, mapping hidden states to vocabulary-sized logits, often constitutes a substantial portion of the model's parameters and computational cost during inference. Existing methods like adaptive softmax or hierarchical softmax introduce structural complexities. In this paper, we propose VQ-Logits, a novel approach that leverages Vector Quantization (VQ) to drastically reduce the parameter count and computational load of the LLM output layer. VQ-Logits replaces the large V * dmodel output embedding matrix with a small, shared codebook of K embedding vectors (K << V ). Each token in the vocabulary is mapped to one of these K codebook vectors. The LLM predicts logits over this compact codebook, which are then efficiently "scattered" to the full vocabulary space using the learned or preassigned mapping. We demonstrate through extensive experiments on standard language modeling benchmarks (e.g., WikiText-103, C4) that VQ-Logits can achieve up to 99% parameter reduction in the output layer and 6x speedup in logit computation, with only a marginal 4% increase in perplexity compared to full softmax baselines. We further provide detailed ablation studies on codebook size, initialization, and learning strategies, showcasing the robustness and effectiveness of our approach.

cs.CL↗

arXiv2025

ComplexFormer: Disruptively Advancing Transformer Inference Ability via Head-Specific Complex Vector Attention

Transformer models rely on self-attention to capture token dependencies but face challenges in effectively integrating positional information while allowing multi-head attention (MHA) flexibility. Prior methods often model semantic and positional differences disparately or apply uniform positional adjustments across heads, potentially limiting representational capacity. This paper introduces ComplexFormer, featuring Complex Multi-Head Attention-CMHA. CMHA empowers each head to independently model semantic and positional differences unified within the complex plane, representing interactions as rotations and scaling. ComplexFormer incorporates two key improvements: (1) a per-head Euler transformation, converting real-valued query/key projections into polar-form complex vectors for head-specific complex subspace operation; and (2) a per-head adaptive differential rotation mechanism, exp[i(Adapt(ASmn,i) + Delta(Pmn),i)], allowing each head to learn distinct strategies for integrating semantic angle differences (ASmn,i) with relative positional encodings (Delta(Pmn),i). Extensive experiments on language modeling, text generation, code generation, and mathematical reasoning show ComplexFormer achieves superior performance, significantly lower generation perplexity , and improved long-context coherence compared to strong baselines like RoPE-Transformers. ComplexFormer demonstrates strong parameter efficiency, offering a more expressive, adaptable attention mechanism.

cs.LG↗

arXiv2023

Effects of a rotating periodic lattice on coherent quantum states in a ring topology: The case of positive nonlinearity

We study the landscape of solutions of the coherent quantum states in a ring shaped lattice potential in the context of ultracold atoms with an effective positive nonlinearity induced by interatomic interactions. The exact analytical solutions in the absence of lattice are used as a starting point and the transformation of those solutions is mapped as the lattice is introduced and strengthened. This approach allows a simple classification of all the solutions into states with periods commensurate/incommensruate with the lattice period and those with/without nodes. Their origins are traced to the primary dispersion curve and the swallowtail branches of the lattice-free spectrum. The commensurate states tend to remain delocalized with increasing lattice depth, whereas the incommensurate ones may be localized. The symmetry and stability properties of the solutions are examined and correlated with branch energies. The crucial importance of rotation is highlighted by its utility in continuously transforming solutions and accessing in a finite ring with a few sites the full spectrum of nonlinear Bloch waves on an infinite lattice.

cond-mat.quant-gas↗

arXiv2023

Quantum Scattering States in a Nonlinear Coherent Medium

We present a comprehensive study of stationary states in a coherent medium with a quadratic or Kerr nonlinearity in the presence of localized potentials in one dimension (1D) for both positive and negative signs of the nonlinear term, as well as for barriers and wells. The description is in terms of the nonlinear Schrödinger equation (NLSE) and hence applicable to a variety of systems, including interacting ultracold atoms in the mean field regime and light propagation in optical fibers. We determine the full landscape of solutions, in terms of a potential step and build solutions for rectangular barrier and well potentials. It is shown that all the solutions can be expressed in terms of a Jacobi elliptic function with the inclusion of a complex-valued phase shift. Our solution method relies on the roots of a cubic polynomial associated with a hydrodynamic picture, which provides a simple classification of all the solutions, both bounded and unbounded, while the boundary conditions are intuitively visualized as intersections of phase space curves. We compare solutions for open boundary conditions with those for a barrier potential on a ring, and also show that numerically computed solutions for smooth barriers agree qualitatively with analytical solutions for rectangular barriers. A stability analysis of solutions based on the Bogoliubov equations for fluctuations show that persistent instabilities are localized at sharp boundaries, and are predicated by the relation of the mean density change across the boundary to the value of the derivative of the density at the edge. We examine the scattering of a wavepacket by a barrier potential and show that at any instant the scattered states are well described by the stationary solutions we obtain, indicating applications of our results and methods to nonlinear scattering problems.

cond-mat.quant-gas↗

arXiv2022

X-PuDu at SemEval-2022 Task 6: Multilingual Learning for English and Arabic Sarcasm Detection

Detecting sarcasm and verbal irony from people's subjective statements is crucial to understanding their intended meanings and real sentiments and positions in social scenarios. This paper describes the X-PuDu system that participated in SemEval-2022 Task 6, iSarcasmEval - Intended Sarcasm Detection in English and Arabic, which aims at detecting intended sarcasm in various settings of natural language understanding. Our solution finetunes pre-trained language models, such as ERNIE-M and DeBERTa, under the multilingual settings to recognize the irony from Arabic and English texts. Our system ranked second out of 43, and ninth out of 32 in Task A: one-sentence detection in English and Arabic; fifth out of 22 in Task B: binary multi-label classification in English; first out of 16, and fifth out of 13 in Task C: sentence-pair detection in English and Arabic.

cs.CL↗

arXiv2019

Structured Neural Topic Models for Reviews

We present Variational Aspect-based Latent Topic Allocation (VALTA), a family of autoencoding topic models that learn aspect-based representations of reviews. VALTA defines a user-item encoder that maps bag-of-words vectors for combined reviews associated with each paired user and item onto structured embeddings, which in turn define per-aspect topic weights. We model individual reviews in a structured manner by inferring an aspect assignment for each sentence in a given review, where the per-aspect topic weights obtained by the user-item encoder serve to define a mixture over topics, conditioned on the aspect. The result is an autoencoding neural topic model for reviews, which can be trained in a fully unsupervised manner to learn topics that are structured into aspects. Experimental evaluation on large number of datasets demonstrates that aspects are interpretable, yield higher coherence scores than non-structured autoencoding topic model variants, and can be utilized to perform aspect-based comparison and genre discovery.

cs.CL↗

arXiv2018

Reimplementation and Reinterpretation of the Copycat Project

We present the reinterpreted and reimplemented Copycat project, an architecture solving letter analogy domain problems. To support a flexible implementation change and rigor testing process, we propose a implementation method in DrRacket by using functional abstraction, naming system, initialization, and structural reference. Finally, benefits and limitations are analyzed for cognitive architectures along the lines of Copycat.

cs.AI↗