Large Language Models (LLMs) have shown remarkable reasoning capabilities on complex tasks, but they still suffer from out-of-date knowledge, hallucinations, and opaque decision-making. In contrast, Knowledge Graphs (KGs) can provide explicit and editable knowledge for LLMs to alleviate these issues. Existing paradigm of KG-augmented LLM manually predefines the breadth of exploration space and requires flawless navigation in KGs. However, this paradigm cannot adaptively explore reasoning paths in KGs based on the question semantics and self-correct erroneous reasoning paths, resulting in a bottleneck in efficiency and effect. To address these limitations, we propose a novel self-correcting adaptive planning paradigm for KG-augmented LLM named Plan-on-Graph (PoG), which first decomposes the question into several sub-objectives and then repeats the process of adaptively exploring reasoning paths, updating memory, and reflecting on the need to self-correct erroneous reasoning paths until arriving at the answer. Specifically, three important mechanisms of Guidance, Memory, and Reflection are designed to work together, to guarantee the adaptive breadth of self-correcting planning for graph reasoning. Finally, extensive experiments on three real-world datasets demonstrate the effectiveness and efficiency of PoG.

This paper proposes a novel reinforcement learning framework to address the Liner Shipping Network Design Problem (LSNDP), a challenging combinatorial optimization problem focused on designing cost-efficient maritime shipping routes. Traditional methods for solving the LSNDP typically involve decomposing the problem into sub-problems, such as network design and multi-commodity flow, which are then tackled using approximate heuristics or large neighborhood search (LNS) techniques. In contrast, our approach employs a model-free reinforcement learning algorithm on the network design, integrated with a heuristic-based multi-commodity flow solver, to produce competitive results on the publicly available LINERLIB benchmark. Additionally, our method also demonstrates generalization capabilities by producing competitive solutions on the benchmark instances after training on perturbed instances.

Large Language Models (LLMs) have shown remarkable reasoning capabilities on complex tasks, but they still suffer from out-of-date knowledge, hallucinations, and opaque decision-making. In contrast, Knowledge Graphs (KGs) can provide explicit and editable knowledge for LLMs to alleviate these issues. Existing paradigm of KG-augmented LLM manually predefines the breadth of exploration space and requires flawless navigation in KGs. However, this paradigm cannot adaptively explore reasoning paths in KGs based on the question semantics and self-correct erroneous reasoning paths, resulting in a bottleneck in efficiency and effect. To address these limitations, we propose a novel self-correcting adaptive planning paradigm for KG-augmented LLM named Plan-on-Graph (PoG), which first decomposes the question into several sub-objectives and then repeats the process of adaptively exploring reasoning paths, updating memory, and reflecting on the need to self-correct erroneous reasoning paths until arriving at the answer. Specifically, three important mechanisms of Guidance, Memory, and Reflection are designed to work together, to guarantee the adaptive breadth of self-correcting planning for graph reasoning. Finally, extensive experiments on three real-world datasets demonstrate the effectiveness and efficiency of PoG.

Accurate forecasting of electrical demand is essential for maintaining a stable and reliable power grid, optimizing the allocation of energy resources, and promoting efficient energy consumption practices. This study investigates the effectiveness of five hyperparameter optimization (HPO) algorithms -- Random Search, Covariance Matrix Adaptation Evolution Strategy (CMA--ES), Bayesian Optimization, Partial Swarm Optimization (PSO), and Nevergrad Optimizer (NGOpt) across univariate and multivariate Short-Term Load Forecasting (STLF) tasks. Using the Panama Electricity dataset (n=48,049), we evaluate HPO algorithms' performances on a surrogate forecasting algorithm, XGBoost, in terms of accuracy (i.e., MAPE, $R^2$) and runtime. Performance plots visualize these metrics across varying sample sizes from 1,000 to 20,000, and Kruskal--Wallis tests assess the statistical significance of the performance differences. Results reveal significant runtime advantages for HPO algorithms over Random Search. In univariate models, Bayesian optimization exhibited the lowest accuracy among the tested methods. This study provides valuable insights for optimizing XGBoost in the STLF context and identifies areas for future research.

The computer scientist and CEO who popularized the term'artificial general intelligence' (AGI) believes AI is verging on an exponential'intelligence explosion.' The PhD mathematician and futurist Ben Goertzel made the prediction while closing out a summit on AGI this month: 'It seems quite plausible we could get to human-level AGI within, let's say, the next three to eight years.' 'Once you get to human-level AGI,' Goertzel, sometimes called'father of AGI,' added, 'within a few years you could get a radically superhuman AGI.' While the futurist admitted that he'could be wrong,' he went on to predict that the only impediment to a runaway, ultra-advanced AI -- far more advanced than its human makers -- would be if the bot's'own conservatism' advised caution. Mathematician and futurist Ben Goertzel made the prediction while closing out a summit on AGI las week: 'we could get to human-level AGI within, let's say, the next three to eight years' Goertzel made his predictions during his closing remarks last week at the '2024 Beneficial AI Summit and Unconference,' partially sponsored by his own firm SingularityNET where he is CEO.

Google wants your help in preserving and restoring coral reefs, and has designed a platform to help with this mission in mere minutes. All you have to do is tune in. Called "Calling in Our Corals"(opens in a new tab), the new citizen science project is a collaboration between Google Arts and Culture and marine biologists across the globe. People are being asked to listen to underwater recordings of coral reefs in Marine Protected Areas through an online platform, identifying sounds made by fish, shrimp, and other marine creatures in order to monitor ecosystems and determine opportunities for reef restoration. Scientists have placed hydrophones in 10 reefs across Australia, Indonesia, Philippines, the U.S., Panama, and Sweden, which record 24 hours a day, meaning there's hundreds of hours to sift through.

The Mayflower Autonomous Ship finally arrived on the coast of Nova Scotia last month, marking the end of its long trek across the Atlantic. While the modern Mayflower is far from the first vessel to make that voyage, this small robotic boat is the largest to ever do so navigated by artificial intelligence with no humans aboard. A few technical hiccups notwithstanding, its trip is the latest evidence that the future of the high seas could be autonomous. Slowly, self-steering ships are becoming a reality. In Norway, an autonomous battery-powered container vessel is shuttling fertilizer between a factory and a local port, and pending a successful trial, it could be fully certified within the next two years.

Avikus, a subsidiary of Hyundai, has successfully completed the first transoceanic voyage of a large merchant ship using autonomous navigation technologies, the company said in a press release. With the increase in computational capabilities, autonomous navigation solutions are being tested in various fields of transportation. Autonomous cars are expected to bring in a new era of human transportation, and the maritime industry is also not very far behind. Last year, we reported on a fertilizer company that deployed a fully electric and autonomous container ship in Norway to save 40,000 truck trips every year. While this deployment was over a short distance, maritime transportation involves crossing oceans and often in very congested port areas.

As a result, companies have gone through a decade's worth of digital transformation in just a matter of months, with the pandemic forcing them to refresh archaic processes with AI, machine learning, and data science technologies. Such technological advancements will continue to evolve and further establish themselves as a critical component to managing complex logistical landscapes – from improving efficiency and mitigating the effects of a global labour shortage, to identifying more robust and dependable ways to move commodities. In a world where uncertainty is the only certainty, AI-enabled order and inventory visibility across shipments will also be vital to'keep the wheels in motion.' Most importantly, to provide real-time updates on changes to arrival times and to identify potential disruptions before and as they occur. Take the recent congestion issues at the Port of Los Angeles, for example.

Using artificial intelligence for rapid data collection and integration of shrunk the commander's decision cycle from days to minutes in some instances in a recent information experiment by U.S. Northern Command, the head of NORTHCOM said Wednesday. Speaking to reporters at the Pentagon, Gen. Glen VanHerck said the Global Information Dominance Exercise or GIDE, "focused a lot on contested logistics to give us a scenario where maybe a line of communication such as the Panama Canal may be challenged," by a peer competitor such as China or Russia. The experiment wrapped up during the second week of July. The experiment was hosted by NORTHCOM but included 11 combatant commands, which illustrated how they can integrate and act on data from satellites, planes, and other sources. It also tested the command's ability to use new artificial intelligence abilities to monitor and predict potential threats using those data sources.