Enhancing AI's Reasoning with Active Retrieval and Monte Carlo Tree Search
A clear explanation of AR-MCTS, a framework that improves multimodal reasoning by combining active retrieval, Monte Carlo Tree Search, and process rewards.
Blog Topic
Neurosymbolic & Trustworthy AI Articles
Research notes on neurosymbolic learning, symbolic reasoning, knowledge distillation, and explainable AI.
Demystifying the Difference Between Explainable AI and Neurosymbolic AI
A practical explanation of how Explainable AI and Neurosymbolic AI differ, where they overlap, and why both matter for trustworthy systems.
Symbolic Knowledge Distillation of Large Language Models
A guide to symbolic knowledge distillation for LLMs, including direct, multilevel, and reinforcement-learning approaches.
Giving AI Some Common Sense
A readable reflection on Ron Brachman's Northeastern lecture about why AI systems need common sense before they can be trusted to act autonomously.
Predictions for abundant, awesome, technology-based, Possible Tomorrows (2035-2049)
A more grounded look at Vinod Khosla's TED 2024 vision for technology-driven possible tomorrows between 2035 and 2049.
Interval Temporal Logic Decision Tree Learning
A readable explanation of Temporal ID3, a decision-tree learning method that uses interval temporal logic to classify timelines instead of static records.
Towards a General Method for Logical Rule Extraction from Time Series
A readable explanation of a general pipeline for turning multivariate time series into timelines and extracting temporal logical rules from them.
A Brief Overview of Rule Learning
An overview of descriptive and predictive rule learning, association rules, decision lists, and covering algorithms.
Symbolic regression outperforms other models for small data sets
A practical look at why symbolic regression can be a strong choice when data is scarce, interpretability matters, and overfitting is a constant risk.
Deep Symbolic Regression for Recurrent Sequences
A readable explanation of how Transformers can infer symbolic recurrence relations from integer and float sequences, then use those formulas to extrapolate.
Learning Symbolic Physics with Graph Networks
A readable explanation of how graph networks can learn physical interactions, expose force-like messages, and support symbolic law discovery.
Distilling Free-Form Natural Laws from Experimental Data
A readable look at how symbolic regression can rediscover physical laws directly from experimental data, from simple oscillators to chaotic double pendulums.
Discovering Symbolic Models from Deep Learning with Inductive Biases
Learn how inductive biases, sparse latent spaces, graph networks, and symbolic regression extract equations from deep models.
Symbolic Regression Enhanced Decision Tree
Learn how symbolic regression creates expressive split rules for compact, interpretable decision-tree classifiers.
Knowledge Representation and Knowledge Embeddings in Neurosymbolic AI
A practical explanation of symbolic knowledge representations and where that knowledge can be embedded inside neuro-symbolic AI systems.
Neurosymbolic Approaches in Artificial Intelligence
Review major neuro-symbolic AI approaches that connect deep learning, symbolic knowledge, explainability, and hybrid reasoning.
Approaches for Explainable Artificial Intelligence
Learn about the RVS framework for explainable AI, including feature, design, representation, training-data, and stakeholder explanations.
Neurosymbolic AI and Trustworthiness
Explore how neuro-symbolic AI can improve interpretability, fairness, privacy, safety, and trustworthy reasoning.
Decision Trees with Short Explainable Rules
Learn how decision trees can optimize for shorter explanations by reducing the number of distinct attributes used in prediction rules.
From Generative AI to Trustworthy AI
Explore what generative AI needs to become trustworthy, including reasoning, provenance, ethics, context, and symbolic knowledge.
Neurosymbolic Reinforcement Learning: An Overview
Learn how neurosymbolic reinforcement learning combines neural learning, symbolic reasoning, and planning for interpretable and safer agents.
Explainable AI in Cybersecurity
Explore explainable AI in cybersecurity for intrusion detection, malware analysis, anomaly detection, risk scoring, debugging, and model trust.
Neurosymbolic AI
An introduction to neurosymbolic AI and how neural networks and symbolic reasoning support explainability, data efficiency, and trustworthy decisions.