100x Cost & Latency Reduction: Performance Analysis of AI Query Approximation using Lightweight Proxy Models

Several data warehouse and database providers have recently introduced extensions to SQL called AI Queries, enabling users to specify functions and conditions in SQL that are evaluated by LLMs, thereby broadening significantly the kinds of queries one can express over the combination of structured and unstructured data. LLMs offer remarkable semantic reasoning capabilities, making them an essential tool for complex and nuanced queries that blend structured and unstructured data. While extremely powerful, these AI queries can become prohibitively costly when invoked thousands of times. This paper provides an extensive evaluation of a recent AI query approximation approach that enables low cost analytics and database applications to benefit from AI queries. The approach delivers >100x cost and latency reduction for the semantic filter (AI.IF) operator and also important gains for semantic ranking (AI.RANK). The cost and performance gains come from utilizing cheap and accurate proxy models over embedding vectors. We show that despite the massive gains in latency and cost, these proxy models preserve accuracy and occasionally improve accuracy across various benchmark datasets, including the extended Amazon reviews benchmark that has 10M rows. We present an OLAP-friendly architecture within Google \textit{BigQuery} for this approach for purely online (ad hoc) queries, and a low-latency HTAP database-friendly architecture in \textit{AlloyDB} that could further improve the latency by moving the proxy model training offline. We present techniques that accelerate the proxy model training.

NanoVDR: Distilling a 2B Vision-Language Retriever into a 70M Text-Only Encoder for Visual Document Retrieval

Vision-Language Model (VLM) based retrievers have advanced visual document retrieval (VDR) to impressive quality. They require the same multi-billion parameter encoder for both document indexing and query encoding, incurring high latency and GPU dependence even for plain-text queries. We observe that this design is unnecessarily symmetric: documents are visually complex and demand strong visual understanding, whereas queries are just short text strings. NanoVDR exploits this query--document asymmetry by decoupling the two encoding paths: a frozen 2B VLM teacher indexes documents offline, while a distilled text-only student as small as 69M parameters encodes queries at inference. The key design choice is the distillation objective. Through systematic comparison of six objectives across three backbones and 22 ViDoRe benchmark datasets, we find that pointwise cosine alignment on query text consistently outperforms ranking-based and contrastive alternatives, while requiring only pre-cached teacher query embeddings and no document processing during training. Furthermore, we identify cross-lingual transfer as the primary performance bottleneck, and resolve it cheaply by augmenting training data with machine-translated queries. The resulting NanoVDR-S-Multi (DistilBERT, 69M) retains 95.1\% of teacher quality and outperforms DSE-Qwen2 (2B) on v2 and v3 with 32$\times$ fewer parameters and 50$\times$ lower CPU query latency, at a total training cost under 13 GPU-hours.

LingoMotion: An Interpretable and Unambiguous Symbolic Representation for Human Motion

Existing representations for human motion, such as MotionGPT, often operate as black-box latent vectors with limited interpretability and build on joint positions which can cause ambiguity. Inspired by the hierarchical structure of natural languages - from letters to words, phrases, and sentences - we propose LingoMotion, a motion language that facilitates interpretable and unambiguous symbolic representation for both simple and complex human motion. In this paper, we introduce the concept design of LingoMotion, including the definitions of motion alphabet based on joint angles, the morphology for forming words and phrases to describe simple actions like walking and their attributes like speed and scale, as well as the syntax for describing more complex human activities with sequences of words and phrases. The preliminary results, including the implementation and evaluation of motion alphabet using a large-scale motion dataset Motion-X, demonstrate the high fidelity of motion representation.

Fine-grained Motion Retrieval via Joint-Angle Motion Images and Token-Patch Late Interaction

Text-motion retrieval aims to learn a semantically aligned latent space between natural language descriptions and 3D human motion skeleton sequences, enabling bidirectional search across the two modalities. Most existing methods use a dual-encoder framework that compresses motion and text into global embeddings, discarding fine-grained local correspondences, and thus reducing accuracy. Additionally, these global-embedding methods offer limited interpretability of the retrieval results. To overcome these limitations, we propose an interpretable, joint-angle-based motion representation that maps joint-level local features into a structured pseudo-image, compatible with pre-trained Vision Transformers. For text-to-motion retrieval, we employ MaxSim, a token-wise late interaction mechanism, and enhance it with Masked Language Modeling regularization to foster robust, interpretable text-motion alignment. Extensive experiments on HumanML3D and KIT-ML show that our method outperforms state-of-the-art text-motion retrieval approaches while offering interpretable fine-grained correspondences between text and motion. The code is available in the supplementary material.

Look in the Middle: Structural Anchor Pruning for Scalable Visual RAG Indexing

Recent Vision-Language Models (e.g., ColPali) enable fine-grained Visual Document Retrieval (VDR) but incur prohibitive index vector size overheads. Training-free pruning solutions (e.g., EOS-attention based methods) can reduce index vector size by approximately 60% without model adaptation, but often underperform random selection in high-compression scenarios (> 80%). Prior research (e.g., Light-ColPali) attributes this to the conclusion that visual token importance is inherently query-dependent, thereby questioning the feasibility of training-free pruning. In this work, we propose Structural Anchor Pruning (SAP), a training-free pruning method that identifies key visual patches from middle layers to achieve high performance compression. We also introduce Oracle Score Retention (OSR) protocol to evaluate how layer-wise information affects compression efficiency. Evaluations on the ViDoRe benchmark demonstrate that SAP reduces index vectors by over 90% while maintaining robust retrieval fidelity, providing a highly scalable solution for Visual RAG. Furthermore, our OSR-based analysis reveals that semantic structural anchor patches persist in the middle layers, unlike traditional pruning solutions that focus on the final layer where structural signals dissipate.

ColorConceptBench: A Benchmark for Probabilistic Color-Concept Understanding in Text-to-Image Models

While text-to-image (T2I) models have advanced considerably, their capability to associate colors with implicit concepts remains underexplored. To address the gap, we introduce ColorConceptBench, a new human-annotated benchmark to systematically evaluate color-concept associations through the lens of probabilistic color distributions. ColorConceptBench moves beyond explicit color names or codes by probing how models translate 1,281 implicit color concepts using a foundation of 6,369 human annotations. Our evaluation of seven leading T2I models reveals that current models lack sensitivity to abstract semantics, and crucially, this limitation appears resistant to standard interventions (e.g., scaling and guidance). This demonstrates that achieving human-like color semantics requires more than larger models, but demands a fundamental shift in how models learn and represent implicit meaning.

Zero-shot Load Forecasting for Integrated Energy Systems: A Large Language Model-based Framework with Multi-task Learning

The growing penetration of renewable energy sources in power systems has increased the complexity and uncertainty of load forecasting, especially for integrated energy systems with multiple energy carriers. Traditional forecasting methods heavily rely on historical data and exhibit limited transferability across different scenarios, posing significant challenges for emerging applications in smart grids and energy internet. This paper proposes the TSLLM-Load Forecasting Mechanism, a novel zero-shot load forecasting framework based on large language models (LLMs) to address these challenges. The framework consists of three key components: a data preprocessing module that handles multi-source energy load data, a time series prompt generation module that bridges the semantic gap between energy data and LLMs through multi-task learning and similarity alignment, and a prediction module that leverages pre-trained LLMs for accurate forecasting. The framework's effectiveness was validated on a real-world dataset comprising load profiles from 20 Australian solar-powered households, demonstrating superior performance in both conventional and zero-shot scenarios. In conventional testing, our method achieved a Mean Squared Error (MSE) of 0.4163 and a Mean Absolute Error (MAE) of 0.3760, outperforming existing approaches by at least 8\%. In zero-shot prediction experiments across 19 households, the framework maintained consistent accuracy with a total MSE of 11.2712 and MAE of 7.6709, showing at least 12\% improvement over current methods. The results validate the framework's potential for accurate and transferable load forecasting in integrated energy systems, particularly beneficial for renewable energy integration and smart grid applications.

Exploring the Inquiry-Diagnosis Relationship with Advanced Patient Simulators

Online medical consultation (OMC) restricts doctors to gathering patient information solely through inquiries, making the already complex sequential decision-making process of diagnosis even more challenging. Recently, the rapid advancement of large language models has demonstrated a significant potential to transform OMC. However, most studies have primarily focused on improving diagnostic accuracy under conditions of relatively sufficient information, while paying limited attention to the "inquiry" phase of the consultation process. This lack of focus has left the relationship between "inquiry" and "diagnosis" insufficiently explored. In this paper, we first extract real patient interaction strategies from authentic doctor-patient conversations and use these strategies to guide the training of a patient simulator that closely mirrors real-world behavior. By inputting medical records into our patient simulator to simulate patient responses, we conduct extensive experiments to explore the relationship between "inquiry" and "diagnosis" in the consultation process. Experimental results demonstrate that inquiry and diagnosis adhere to the Liebig's law: poor inquiry quality limits the effectiveness of diagnosis, regardless of diagnostic capability, and vice versa. Furthermore, the experiments reveal significant differences in the inquiry performance of various models. To investigate this phenomenon, we categorize the inquiry process into four types: (1) chief complaint inquiry; (2) specification of known symptoms; (3) inquiry about accompanying symptoms; and (4) gathering family or medical history. We analyze the distribution of inquiries across the four types for different models to explore the reasons behind their significant performance differences. We plan to open-source the weights and related code of our patient simulator at https://github.com/LIO-H-ZEN/PatientSimulator.

Transformer-Based Approaches for Sensor-Based Human Activity Recognition: Opportunities and Challenges

Transformers have excelled in natural language processing and computer vision, paving their way to sensor-based Human Activity Recognition (HAR). Previous studies show that transformers outperform their counterparts exclusively when they harness abundant data or employ compute-intensive optimization algorithms. However, neither of these scenarios is viable in sensor-based HAR due to the scarcity of data in this field and the frequent need to perform training and inference on resource-constrained devices. Our extensive investigation into various implementations of transformer-based versus non-transformer-based HAR using wearable sensors, encompassing more than 500 experiments, corroborates these concerns. We observe that transformer-based solutions pose higher computational demands, consistently yield inferior performance, and experience significant performance degradation when quantized to accommodate resource-constrained devices. Additionally, transformers demonstrate lower robustness to adversarial attacks, posing a potential threat to user trust in HAR.

Enhancing Motion Variation in Text-to-Motion Models via Pose and Video Conditioned Editing

Text-to-motion models that generate sequences of human poses from textual descriptions are garnering significant attention. However, due to data scarcity, the range of motions these models can produce is still limited. For instance, current text-to-motion models cannot generate a motion of kicking a football with the instep of the foot, since the training data only includes martial arts kicks. We propose a novel method that uses short video clips or images as conditions to modify existing basic motions. In this approach, the model's understanding of a kick serves as the prior, while the video or image of a football kick acts as the posterior, enabling the generation of the desired motion. By incorporating these additional modalities as conditions, our method can create motions not present in the training set, overcoming the limitations of text-motion datasets. A user study with 26 participants demonstrated that our approach produces unseen motions with realism comparable to commonly represented motions in text-motion datasets (e.g., HumanML3D), such as walking, running, squatting, and kicking.