Conditional Density Estimation via Least-Squares Density Ratio Estimation

⋅

Estimating the conditional mean of an input-output relation is the goal of regression. However, regression analysis is not sufficiently informative if the conditional distribution has multi-modality, is highly asymmetric, or contains heteroscedastic noise. In such scenarios, estimating the conditional distribution itself would be more useful. In this paper, we propose a novel method of conditional density estimation that is suitable for multi-dimensional continuous variables. The basic idea of the proposed method is to express the conditional density in […]

Ver mais

Liked Liked

Collaborative and Efficient Fine-tuning: Leveraging Task Similarity

⋅

arXiv:2602.07218v1 Announce Type: cross Abstract: Adaptability has been regarded as a central feature in the foundation models, enabling them to effectively acclimate to unseen downstream tasks. Parameter-efficient fine-tuning methods such as celebrated LoRA facilitate efficient adaptation of large foundation models using labeled, high-quality and generally scarce task data. To mitigate data scarcity in fine-tuning of foundation models, we propose to leverage task similarity across multiple downstream users. Intuitively, users with similar tasks must be able to assist each […]

Ver mais

Liked Liked

Rethinking LoRA for Data Heterogeneous Federated Learning: Subspace and State Alignment

⋅

Low-Rank Adaptation (LoRA) is widely used for federated fine-tuning. Yet under non-IID settings, it can substantially underperform full-parameter fine-tuning. Through with-high-probability robustness analysis, we uncover that this gap can be attributed to two coupled mismatches: (i) update-space mismatch, where clients optimize in a low-rank subspace but aggregation occurs in the full space; and (ii) optimizer-state mismatch, where unsynchronized adaptive states amplify drift across rounds. We propose FedGaLore, which combines client-side GaLore-style gradient-subspace optimization with server-side drift-robust synchronization of […]

Ver mais

Liked Liked

Synthesizing the Virtual Advocate: A Multi-Persona Speech Generation Framework for Diverse Linguistic Jurisdictions in Indic Languages

⋅

arXiv:2602.11172v1 Announce Type: new Abstract: Legal advocacy requires a unique combination of authoritative tone, rhythmic pausing for emphasis, and emotional intelligence. This study investigates the performance of the Gemini 2.5 Flash TTS and Gemini 2.5 Pro TTS models in generating synthetic courtroom speeches across five Indic languages: Tamil, Telugu, Bengali, Hindi, and Gujarati. We propose a prompting framework that utilizes Gemini 2.5s native support for 5 languages and its context-aware pacing to produce distinct advocate personas. The evolution […]

Ver mais

Liked Liked

[R] IDA PhD Forum CfP (deadline Feb 23), get feedback and mentorship on your research

⋅

Calling all AI/ML PhD students out there, get feedback on your research plus mentorship from senior researchers at the 2026 Symposium on Intelligent Data Analysis. 2 page abstract deadline Feb 23, 2026. Call for papers Leiden (Netherlands) April 22-24, 2026 (Wednesday – Friday) https://ida2026.liacs.nl/ IDA is organizing the 2026 edition of the PhD Forum, aimed at PhD students. This mentoring program aims to connect PhD students with senior scientists who share their experience to help advance the students’ […]

Ver mais

Liked Liked

FROST: Filtering Reasoning Outliers with Attention for Efficient Reasoning

⋅

arXiv:2601.19001v1 Announce Type: new Abstract: We propose FROST, an attention-aware method for efficient reasoning. Unlike traditional approaches, FROST leverages attention weights to prune uncritical reasoning paths, yielding shorter and more reliable reasoning trajectories. Methodologically, we introduce the concept of reasoning outliers and design an attention-based mechanism to remove them. Theoretically, FROST preserves and enhances the model’s reasoning capacity while eliminating outliers at the sentence level. Empirically, we validate FROST on four benchmarks using two strong reasoning models (Phi-4-Reasoning […]

Ver mais

Liked Liked
Struggling to get PPO to work for pickup & delivery task — stuck, need for guidance

Struggling to get PPO to work for pickup & delivery task — stuck, need for guidance

⋅

Hi everyone, we’re a group of students working on a reinforcement learning project and we’re honestly pretty stuck. We’ve been trying to solve this for weeks and feel like we’re missing some fundamental understanding. Below are the main problems we’re facing: Problem Setup We train an agent to pick up items and deliver them to a target on a small grid. We use PPO as we thought it is most advanced We split the reward 50/50 between pickup […]

Ver mais

Liked Liked

Beyond Interaction Effects: Two Logics for Studying Population Inequalities

⋅

arXiv:2601.04223v1 Announce Type: new Abstract: When sociologists and other social scientist ask whether the return to college differs by race and gender, they face a choice between two fundamentally different modes of inquiry. Traditional interaction models follow deductive logic: the researcher specifies which variables moderate effects and tests these hypotheses. Machine learning methods follow inductive logic: algorithms search across vast combinatorial spaces to discover patterns of heterogeneity. This article develops a framework for navigating between these approaches. We […]

Ver mais

Liked Liked
I trained an AI to navigate through asteroids in Godot 4.6 using reinforcement learning

I trained an AI to navigate through asteroids in Godot 4.6 using reinforcement learning

⋅

Hey! Been working on this in the past two months. The AI (Rookie) learns to fly through asteroid fields using PPO, no scripted movement, just raw thrust/rotation inputs and a reward system. Everything built in Godot 4.6, models in Blender. I’ve experimented with RL in Godot before, but this is the first time I actually got it to work well enough to be worth showing. The reward shaping process was so fun and interesting that it inspired me […]

Ver mais

Liked Liked

Load More

↑

Digitado © 2025