Lifelong learning is essential for intelligent agents operating in dynamic environments. Current large language model (LLM)-based agents, however, remain stateless and unable to accumulate or transfer knowledge over time. Existing benchmarks treat agents as static systems and fail to evaluate lifelong learning capabilities. We present LifelongAgentBench, the first unified benchmark designed to systematically assess the lifelong learning ability of LLM agents. It provides skill-grounded, interdependent tasks across three interactive environments, Database, Operating System, and Knowledge Graph, with automatic label verification, reproducibility, and modular extensibility. Extensive experiments reveal that conventional experience replay has limited effectiveness for LLM agents due to irrelevant information and context length constraints. We further introduce a group self-consistency mechanism that significantly improves lifelong learning performance. We hope LifelongAgentBench will advance the development of adaptive, memory-capable LLM agents.

Neural network-based policies have demonstrated success in many robotic applications, but often lack human-explanability, which poses challenges in safety-critical deployments. To address this, we propose a neuro-symbolic explanation framework that generates a weighted signal temporal logic (wSTL) specification to describe a robot policy in a interpretable form. Existing methods typically produce explanations that are verbose and inconsistent, which hinders explainability, and loose, which do not give meaningful insights into the underlying policy. We address these issues by introducing a simplification process consisting of predicate filtering, regularization, and iterative pruning. We also introduce three novel explainability evaluation metrics -- conciseness, consistency, and strictness -- to assess explanation quality beyond conventional classification metrics. Our method is validated in three simulated robotic environments, where it outperforms baselines in generating concise, consistent, and strict wSTL explanations without sacrificing classification accuracy. This work bridges policy learning with formal methods, contributing to safer and more transparent decision-making in robotics.

Determining and ranking the most salient entities in a text is critical for user-facing systems, especially as users increasingly rely on models to interpret long documents they only partially read. Graded entity salience addresses this need by assigning entities scores that reflect their relative importance in a text. Existing approaches fall into two main categories: subjective judgments of salience, which allow for gradient scoring but lack consistency, and summarization-based methods, which define salience as mention-worthiness in a summary, promoting explainability but limiting outputs to binary labels (entities are either summary-worthy or not). In this paper, we introduce a novel approach for graded entity salience that combines the strengths of both approaches. Using an English dataset spanning 12 spoken and written genres, we collect 5 summaries per document and calculate each entity's salience score based on its presence across these summaries. Our approach shows stronger correlation with scores based on human summaries and alignments, and outperforms existing techniques, including LLMs. We release our data and code at https://github.

The paper presents a new way to analyze multi-task learning algorithms in the online setting by analysing average regret-upper-bound. The authors introduce a two-level algorithm that runs a mirror descent algorithm for each task and adjusts the parameters of the mirror descent before each new task by running another online learning algorithm on the level of tasks. This task-level algorithm is designed to minimize theoretical regret bounds for each task. The authors prove a general bound on the performance of the presented algorithm and show its corollaries for settings with static and dynamic comparators. A version of the algorithm for different task similarity measures is also presented, as well as, an online-to-batch conversion result for learning-to-learn setting with i.i.d.

The original version of this story appeared in Quanta Magazine. When she was 10 years old, Rose Yu got a birthday present that would change her life--and, potentially, the way we study physics. Her uncle got her a computer. That was a rare commodity in China 25 years ago, and the gift did not go unused. At first, Yu mainly played computer games, but in middle school she won an award for web design.

Weaknesses: * From the theoretical analysis, the main weakness might be the analysis on pure continuous features. Nowadays, it is very unlikely to have this scenario in the most challenging machine learning problems. Thus, the theoretical implications can be limited due to this. From the empirical evaluations, I am curious to know why the method was only compared to a very old algorithm such as CART. Is it the only reasonable algorithm out there for decision trees that can be comparable to the method proposed?

If you're starting to feel a little behind in your career because you aren't completely proficient with AI, you don't need to worry. Even beginners can quickly master valuable AI skills without any tech background in the ChatGPT & Automation E-Degree program, and it's on sale right now for just 19.97 This program offers 12 captivating modules that allow you to immerse yourself in more than 25 hours of engaging coursework. It will transform your perception of the digital world. You'll master ChatGPT and over 20 AI tools that are indispensable in facing the dynamic challenges in today's coding, business, and marketing industries.

I've been a freelance journalist for 10 years, usually writing for magazines and websites about cinema. I presented a morning show on Radio Kraków twice a week for about two years. It was only one part of my work, but I really enjoyed it. It was about culture and cinema, and featured a range of people, from artists to activists. I remember interviewing Ukrainians about the Russian invasion for the first programme I presented, back in 2022. I was let go in August 2024, alongside a dozen co-workers who were also part-time. We were told the radio station was having financial problems.

TL;DR: Sellful's all-in-one, white-label business platform lets you build websites, automate tasks, manage clients, and more--now just 349.97 for life (reg. Running an agency or online business shouldn't feel like stitching together 17 different tools and praying they play nice. With Sellful, you can finally ditch the chaos--and do it all from one powerful, AI-powered platform. For a limited time, lifetime access to Sellful's White Label Website Builder & Software ERP Agency Plan is just 349.97 (reg. That's a one-time price for a platform that replaces your CRM, website builder, funnel builder, POS, invoicing system, email marketing tool, project management suite--and about 20 other things you're probably paying monthly for. It's fully white-labeled, too, so you can run everything under your brand or your clients'.

This paper establishes the theoretical foundations of the online scaled gradient methods (OSGM), a framework that utilizes online learning to adapt stepsizes and provably accelerate first-order methods. OSGM quantifies the effectiveness of a stepsize by a feedback function motivated from a convergence measure and uses the feedback to adjust the stepsize through an online learning algorithm. Consequently, instantiations of OSGM achieve convergence rates that are asymptotically no worse than the optimal stepsize. OSGM yields desirable convergence guarantees on smooth convex problems, including 1) trajectory-dependent global convergence on smooth convex objectives; 2) an improved complexity result on smooth strongly convex problems, and 3) local superlinear convergence. Notably, OSGM constitutes a new family of first-order methods with non-asymptotic superlinear convergence, joining the celebrated quasi-Newton methods. Finally, OSGM explains the empirical success of the popular hypergradient-descent heuristic in optimization for machine learning.