arXiv subjects

Quang Dao

Publications and source records attributed to Quang Dao.

4 recordsLinked to original sources

arXiv2026

Weighted Memory Tree: Remembering What Matters for Long-Horizon LLM Agents

Large language model (LLM) agents have demonstrated the ability to solve multi-step tasks requiring planning, tool use, and external information access, yet growing execution histories increase inference cost and expose reasoning to outdated, irrelevant, or misleading information, potentially degrading reasoning quality. Existing memory approaches organize or compress execution histories but provide limited mechanisms for deciding which memories remain active. We introduce the, a hierarchical memory system that organizes execution into tasks, subtasks, and actions while assigning each memory a dynamic retention score. Event-based updates and selection-based decay revise these scores, allowing WMT to preserve useful information, fold completed trajectories, suppress low-utility content, and retain access to folded context. We evaluate WMT on GAIA-Text using Qwen3-8B, Gemma 4 E4B, and Llama-3.1-8B, with ablations and memory-poisoning experiments. Relative to linear memory, WMT improves accuracy by an average of 9.97 percentage points while reducing prompt-token usage by 32.8%. Memory-poisoning experiments show that WMT limits the persistence and propagation of unreliable information. Our results suggest that effective long-horizon agent memory depends less on storing more information than on deciding which information should remain active.

cs.AI↗

arXiv2026

Open, Reliable, and Collective: A Community-Driven Framework for Tool-Using AI Agents

Tool-integrated LLMs retrieve information, perform computations, and take real-world actions, but their reliability depends on both tool-use accuracy and intrinsic tool accuracy, including tool correctness, stability, and safety. While prior work primarily emphasizes tool use, intrinsic tool accuracy remains underexamined. We introduce OpenTools, a community-driven and maintainable toolbox for discovering, using, evaluating, and contributing open-source tools. OpenTools standardizes tool interfaces, converts documented Python functions into reviewable bundles, supports maintainer-triggered evaluation, and combines non-executing risk inspection with optional advisory LLM review. A public web demo allows users to run tools and agents, inspect evidence, contribute tests, and submit tools for maintainer review, while MCP enables controlled access from external applications. Experiments show that community-contributed, task-specific tools yield relative gains of 6% to 22% over an existing toolbox across multiple agent architectures, highlighting the importance of intrinsic tool accuracy.

cs.AI↗

arXiv2024

Lossy Cryptography from Code-Based Assumptions

Over the past few decades, we have seen a proliferation of advanced cryptographic primitives with lossy or homomorphic properties built from various assumptions such as Quadratic Residuosity, Decisional Diffie-Hellman, and Learning with Errors. These primitives imply hard problems in the complexity class $SZK$ (statistical zero-knowledge); as a consequence, they can only be based on assumptions that are broken in $BPP^{SZK}$. This poses a barrier for building advanced primitives from code-based assumptions, as the only known such assumption is Learning Parity with Noise (LPN) with an extremely low noise rate $\frac{\log^2 n}{n}$, which is broken in quasi-polynomial time. In this work, we propose a new code-based assumption: Dense-Sparse LPN, that falls in the complexity class $BPP^{SZK}$ and is conjectured to be secure against subexponential time adversaries. Our assumption is a variant of LPN that is inspired by McEliece's cryptosystem and random $k\mbox{-}$XOR in average-case complexity. We leverage our assumption to build lossy trapdoor functions (Peikert-Waters STOC 08). This gives the first post-quantum alternative to the lattice-based construction in the original paper. Lossy trapdoor functions, being a fundamental cryptographic tool, are known to enable a broad spectrum of both lossy and non-lossy cryptographic primitives; our construction thus implies these primitives in a generic manner. In particular, we achieve collision-resistant hash functions with plausible subexponential security, improving over a prior construction from LPN with noise rate $\frac{\log^2 n}{n}$ that is only quasi-polynomially secure.

cs.CR↗

arXiv2019

Extended Nestohedra and their Face Numbers

Nestohedra are a family of convex polytopes that includes permutohedra, associahedra, and graph associahedra. In this paper, we study an extension of such polytopes, called extended nestohedra. We show that these objects are indeed the boundaries of simple polytopes, answering a question of Lam and Pylyavskyy. We also study the duals of (extended) nestohedra, giving a complete characterization of isomorphisms (as simplicial complexes) between the duals of extended nestohedra and a partial characterization of isomorphisms between the duals of nestohedra and extended nestohedra. In addition, we give formulas for their $f$-, $h$-, and $γ$-vectors. This includes showing that the $f$-vectors of the extended nestohedron corresponding to a forest $F$ and the nestohedron corresponding to the line graph of $F$ are the same, as well as showing that all flag extended nestohedra have nonnegative $γ$-vectors, thus proving Gal's conjecture for a large class of flag simple polytopes. We also relate the $f$- and $h$-vectors of the nestohedra and extended nestohedra, as well as give explicit formulas for the $h$- and $γ$-vectors in terms of descent statistics for a certain class of flag extended nestohedra. Finally, we define a partial ordering on partial permutations that is a join semilattice quotient of the weak Bruhar order on the symmetric group, and such that any linear extension of the partial order provides a shelling of the dual of the stellohedron.

math.CO↗