Repurposing large vision-language models (LVLMs) as computer use agents (CUAs) has led to substantial breakthroughs, primarily driven by human-labeled data. However, these models often struggle with novel and specialized software, particularly in scenarios lacking human annotations. To address this challenge, we propose SEAgent, an agentic self-evolving framework enabling CUAs to autonomously evolve through interactions with unfamiliar software. Specifically, SEAgent empowers computer-use agents to autonomously master novel software environments via experiential learning, where agents explore new software, learn through iterative trial-and-error, and progressively tackle auto-generated tasks organized from simple to complex. To achieve this goal, we design a World State Model for step-wise trajectory assessment, along with a Curriculum Generator that generates increasingly diverse and challenging tasks. The agent's policy is updated through experiential learning, comprised of adversarial imitation of failure actions and Group Relative Policy Optimization (GRPO) on successful ones. Furthermore, we introduce a specialist-to-generalist training strategy that integrates individual experiential insights from specialist agents, facilitating the development of a stronger generalist CUA capable of continuous autonomous evolution. This unified agent ultimately achieves performance surpassing ensembles of individual specialist agents on their specialized software. We validate the effectiveness of SEAgent across five novel software environments within OS-World. Our approach achieves a significant improvement of 23.2% in success rate, from 11.3% to 34.5%, over a competitive open-source CUA, i.e., UI-TARS.

The field of deep learning has witnessed significant breakthroughs, spanning various applications, and fundamentally transforming current software capabilities. However, alongside these advancements, there have been increasing concerns about reproducing the results of these deep learning methods. This is significant because reproducibility is the foundation of reliability and validity in software development, particularly in the rapidly evolving domain of deep learning. The difficulty of reproducibility may arise due to several reasons, including having differences from the original execution environment, incompatible software libraries, proprietary data and source code, lack of transparency, and the stochastic nature in some software. A study conducted by the Nature journal reveals that more than 70% of researchers failed to reproduce other researchers experiments and over 50% failed to reproduce their own experiments. Irreproducibility of deep learning poses significant challenges for researchers and practitioners. To address these concerns, this paper presents a systematic approach at analyzing and improving the reproducibility of deep learning models by demonstrating these guidelines using a case study. We illustrate the patterns and anti-patterns involved with these guidelines for improving the reproducibility of deep learning models. These guidelines encompass establishing a methodology to replicate the original software environment, implementing end-to-end training and testing algorithms, disclosing architectural designs, and enhancing transparency in data processing and training pipelines. We also conduct a sensitivity analysis to understand the model performance across diverse conditions. By implementing these strategies, we aim to bridge the gap between research and practice, so that innovations in deep learning can be effectively reproduced and deployed within software.

Cancer drug response prediction (DRP) models present a promising approach towards precision oncology, tailoring treatments to individual patient profiles. While deep learning (DL) methods have shown great potential in this area, models that can be successfully translated into clinical practice and shed light on the molecular mechanisms underlying treatment response will likely emerge from collaborative research efforts. This highlights the need for reusable and adaptable models that can be improved and tested by the wider scientific community. In this study, we present a scoring system for assessing the reusability of prediction DRP models, and apply it to 17 peer-reviewed DL-based DRP models. As part of the IMPROVE (Innovative Methodologies and New Data for Predictive Oncology Model Evaluation) project, which aims to develop methods for systematic evaluation and comparison DL models across scientific domains, we analyzed these 17 DRP models focusing on three key categories: software environment, code modularity, and data availability and preprocessing. While not the primary focus, we also attempted to reproduce key performance metrics to verify model behavior and adaptability. Our assessment of 17 DRP models reveals both strengths and shortcomings in model reusability. To promote rigorous practices and open-source sharing, we offer recommendations for developing and sharing prediction models. Following these recommendations can address many of the issues identified in this study, improving model reusability without adding significant burdens on researchers. This work offers the first comprehensive assessment of reusability and reproducibility across diverse DRP models, providing insights into current model sharing practices and promoting standards within the DRP and broader AI-enabled scientific research community.

Reproducing published deep learning papers to validate their conclusions can be difficult due to sources of irreproducibility. We investigate the impact that implementation factors have on the results and how they affect reproducibility of deep learning studies. Three deep learning experiments were ran five times each on 13 different hardware environments and four different software environments. The analysis of the 780 combined results showed that there was a greater than 6% accuracy range on the same deterministic examples introduced from hardware or software environment variations alone. To account for these implementation factors, researchers should run their experiments multiple times in different hardware and software environments to verify their conclusions are not affected.

Recently, there has been significant interest in applying machine learning (ML) techniques to X-ray scattering experiments, which proves to be a valuable tool for enhancing research that involves large or rapidly generated datasets. ML allows for the automated interpretation of experimental results, particularly those obtained from synchrotron or neutron facilities. The speed at which ML models can process data presents an important opportunity to establish a closed-loop feedback system, enabling real-time decision-making based on online data analysis. In this study, we describe the incorporation of ML into a closed-loop workflow for X-ray reflectometry (XRR), using the growth of organic thin films as an example. Our focus lies on the beamline integration of ML-based online data analysis and closed-loop feedback. We present solutions that provide an elementary data analysis in real time during the experiment without introducing the additional software dependencies in the beamline control software environment. Our data demonstrates the accuracy and robustness of ML methods for analyzing XRR curves and Bragg reflections and its autonomous control over a vacuum deposition setup.

In computer science, systems are typically divided into two categories: software and hardware. However, there is an additional layer in between, referred to as middleware, which is a software pipeline--an operation, a process, or an application between the operating system and the end user. This article aims to define middleware and reflect on its necessity, as well as address controversies about when and where it applies. It also explores the application of middleware in emerging technologies such as cloud computing and the Internet of Things (IoT), as well as future middleware developments. The term was introduced in the early 1980s.

For modern businesses, gaining value and cutting costs extends to every area of operations--including how they pay their bills. As automation, artificial intelligence, and the digital transformations they support become an increasingly common and important part of effective business process management, automating the accounts payable (AP) function has become a popular first step in achieving greater efficiency, value, and performance. The next step--leveraging artificial intelligence in accounts payable--holds the potential to increase these gains and help organizations of all sizes and types achieve true, end-to-end AP automation. Understanding the importance of advanced artificial intelligence in accounts payable, and adding it to your overall AP optimization paradigm, can help your business meet its goals more effectively while helping to lay a foundation for full digital transformation. Traditionally, AP processes follow the same basic pattern, regardless of industry.

In his recent papers, entitled Intelligence without Representation and Intelligence without Reason, Brooks argues for mobile robots as the foundation of AI research. This article argues that even if we seek to investigate complete agents in real-world environments, robotics is neither necessary nor sufficient as a basis for AI research. The article proposes real-world software environments, such as operating systems or databases, as a complementary substrate for intelligent-agent research and considers the relative advantages of software environments as test beds for AI. First, the cost, effort, and expertise necessary to develop and systematically experiment with software artifacts are relatively low. Second, software environments circumvent many thorny but peripheral research issues that are inescapable in physical environments.

Now that you have answered the questions above, let's move on to the next section. From here on, we would like to draw your attention to some of the most used programming languages for Data Science. You might already be familiar with a few of the popular programming languages, while some may be completely new for you. Python holds a vital place among the top tools for Data Science and is often the go-to choice for a range of tasks for domains such as Machine Learning, Deep Learning, Artificial Intelligence, and more. It is object-oriented, easy to use and extremely developer-friendly thanks to its high code readability.

Largely overlooked in the late 1990s amid the excitement over Y2K, Ray Kurzweil, certified genius, inventor, AI guru, prolific author, and currently director of engineering at Google, started talking about nanobots. While the IT industry was panicking over the prospect of the global failure of millions of computers that had not been programmed to work with a date-field value exceeding 1999, Kurzweil was imagining miniaturized, AI-driven devices that would operate inside human bodies, traveling through the blood, fighting disease, and replenishing cells. With such intelligent robots tending to their bodies' care and maintenance, according to Kurzweil, humans could all expect to live Methuselah-length lifespans. Nanobots come readily to mind because of the loud noises being made of late about robotic process automation (RPA). Surely you have run across the breathless claims about how RPA (sometimes referred to as intelligent process automation or IPA) will help your business save time, save money, improve accuracy, and provide superior customer satisfaction?