SIGMA: Search-Augmented On-Demand Knowledge Integration for Agentic Mathematical Reasoning

Solving mathematical reasoning problems requires not only accurate access to relevant knowledge but also careful, multi-step thinking. However, current retrieval-augmented models often rely on a single perspective, follow inflexible search strategies, and struggle to effectively combine information from multiple sources. We introduce SIGMA (Search-Augmented On-Demand Knowledge Integration for AGentic Mathematical reAsoning), a unified framework that orchestrates specialized agents to independently reason, perform targeted searches, and synthesize findings through a moderator mechanism. Each agent generates hypothetical passages to optimize retrieval for its analytic perspective, ensuring knowledge integration is both context-sensitive and computation-efficient. When evaluated on challenging benchmarks such as MATH500, AIME, and PhD-level science QA GPQA, SIGMA consistently outperforms both open- and closed-source systems, achieving an absolute performance improvement of 7.4%. Our results demonstrate that multi-agent, on-demand knowledge integration significantly enhances both reasoning accuracy and efficiency, offering a scalable approach for complex, knowledge-intensive problem-solving. We will release the code upon publication.

ENTER: Event Based Interpretable Reasoning for VideoQA

In this paper, we present ENTER, an interpretable Video Question Answering (VideoQA) system based on event graphs. Event graphs convert videos into graphical representations, where video events form the nodes and event-event relationships (temporal/causal/hierarchical) form the edges. This structured representation offers many benefits: 1) Interpretable VideoQA via generated code that parses event-graph; 2) Incorporation of contextual visual information in the reasoning process (code generation) via event graphs; 3) Robust VideoQA via Hierarchical Iterative Update of the event graphs. Existing interpretable VideoQA systems are often top-down, disregarding low-level visual information in the reasoning plan generation, and are brittle. While bottom-up approaches produce responses from visual data, they lack interpretability. Experimental results on NExT-QA, IntentQA, and EgoSchema demonstrate that not only does our method outperform existing top-down approaches while obtaining competitive performance against bottom-up approaches, but more importantly, offers superior interpretability and explainability in the reasoning process.

LowCLIP: Adapting the CLIP Model Architecture for Low-Resource Languages in Multimodal Image Retrieval Task

This research explores the development of multimodal vision-language models for image retrieval in low-resource languages, specifically Azerbaijani. Existing vision-language models primarily support high-resource languages, and fine-tuning them remains computationally demanding. To address challenges in vision-language retrieval for low-resource languages, we integrated the CLIP model architecture and employed several techniques to balance computational efficiency with performance. These techniques include synthetic data generation through machine translation, image augmentation, and further training the attention mechanisms of transformer-based models with domain-specific data. We integrated Multilingual BERT as a text encoder with image encoders like ResNet50, EfficientNet0, Vision Transformer (ViT), and Tiny Swin Transformer. Our study found that models like EfficientNet0 and Tiny Swin Transformer perform best on the datasets they were trained on, such as COCO, Flickr30k, and Flickr8k. Augmentation techniques boosted EfficientNet0 MAP on Flickr30k from 0.84 to 0.87 and ResNet50 MAP on MSCOCO from 0.70 to 0.80, contributing to a new state of the art in vision-language retrieval. We share our configurations and results to support further research. Code and pre-trained models are available at https://github.com/aliasgerovs/azclip.

Predicting Financial Market Trends using Time Series Analysis and Natural Language Processing

Forecasting financial market trends through time series analysis and natural language processing poses a complex and demanding undertaking, owing to the numerous variables that can influence stock prices. These variables encompass a spectrum of economic and political occurrences, as well as prevailing public attitudes. Recent research has indicated that the expression of public sentiments on social media platforms such as Twitter may have a noteworthy impact on the determination of stock prices. The objective of this study was to assess the viability of Twitter sentiments as a tool for predicting stock prices of major corporations such as Tesla, Apple. Our study has revealed a robust association between the emotions conveyed in tweets and fluctuations in stock prices. Our findings indicate that positivity, negativity, and subjectivity are the primary determinants of fluctuations in stock prices. The data was analyzed utilizing the Long-Short Term Memory neural network (LSTM) model, which is currently recognized as the leading methodology for predicting stock prices by incorporating Twitter sentiments and historical stock prices data. The models utilized in our study demonstrated a high degree of reliability and yielded precise outcomes for the designated corporations. In summary, this research emphasizes the significance of incorporating public opinions into the prediction of stock prices. The application of Time Series Analysis and Natural Language Processing methodologies can yield significant scientific findings regarding financial market patterns, thereby facilitating informed decision-making among investors. The results of our study indicate that the utilization of Twitter sentiments can serve as a potent instrument for forecasting stock prices, and ought to be factored in when formulating investment strategies.