For decades, human-computer interaction has fundamentally been manual. Even today, almost all productive work done on the computer necessitates human input at every step. Autonomous virtual agents represent an exciting step in automating many of these menial tasks. Virtual agents would empower users with limited technical proficiency to harness the full possibilities of computer systems. They could also enable the efficient streamlining of numerous computer tasks, ranging from calendar management to complex travel bookings, with minimal human intervention. In this paper, we introduce OmniACT, the first-of-a-kind dataset and benchmark for assessing an agent's capability to generate executable programs to accomplish computer tasks. Our scope extends beyond traditional web automation, covering a diverse range of desktop applications. The dataset consists of fundamental tasks such as "Play the next song", as well as longer horizon tasks such as "Send an email to John Doe mentioning the time and place to meet". Specifically, given a pair of screen image and a visually-grounded natural language task, the goal is to generate a script capable of fully executing the task. We run several strong baseline language model agents on our benchmark. The strongest baseline, GPT-4, performs the best on our benchmark However, its performance level still reaches only 15% of the human proficiency in generating executable scripts capable of completing the task, demonstrating the challenge of our task for conventional web agents. Our benchmark provides a platform to measure and evaluate the progress of language model agents in automating computer tasks and motivates future work towards building multimodal models that bridge large language models and the visual grounding of computer screens.

Tokenization is a foundational step in Natural Language Processing (NLP) tasks, bridging raw text and language models. Existing tokenization approaches like Byte-Pair Encoding (BPE) originate from the field of data compression, and it has been suggested that the effectiveness of BPE stems from its ability to condense text into a relatively small number of tokens. We test the hypothesis that fewer tokens lead to better downstream performance by introducing PathPiece, a new tokenizer that segments a document's text into the minimum number of tokens for a given vocabulary. Through extensive experimentation we find this hypothesis not to be the case, casting doubt on the understanding of the reasons for effective tokenization. To examine which other factors play a role, we evaluate design decisions across all three phases of tokenization: pre-tokenization, vocabulary construction, and segmentation, offering new insights into the design of effective tokenizers. Specifically, we illustrate the importance of pre-tokenization and the benefits of using BPE to initialize vocabulary construction. We train 64 language models with varying tokenization, ranging in size from 350M to 2.4B parameters, all of which are made publicly available.

In machine learning (ML), efficient asset management, including ML models, datasets, algorithms, and tools, is vital for resource optimization, consistent performance, and a streamlined development lifecycle. This enables quicker iterations, adaptability, reduced development-to-deployment time, and reliable outputs. Despite existing research, a significant knowledge gap remains in operational challenges like model versioning, data traceability, and collaboration, which are crucial for the success of ML projects. Our study aims to address this gap by analyzing 15,065 posts from developer forums and platforms, employing a mixed-method approach to classify inquiries, extract challenges using BERTopic, and identify solutions through open card sorting and BERTopic clustering. We uncover 133 topics related to asset management challenges, grouped into 16 macro-topics, with software dependency, model deployment, and model training being the most discussed. We also find 79 solution topics, categorized under 18 macro-topics, highlighting software dependency, feature development, and file management as key solutions. This research underscores the need for further exploration of identified pain points and the importance of collaborative efforts across academia, industry, and the research community.

This paper presents FoFo, a pioneering benchmark for evaluating large language models' (LLMs) ability to follow complex, domain-specific formats, a crucial yet underexamined capability for their application as AI agents. Despite LLMs' advancements, existing benchmarks fail to assess their format-following proficiency adequately. FoFo fills this gap with a diverse range of real-world formats and instructions, developed through an AI-Human collaborative method. Our evaluation across both open-source (e.g., Llama 2, WizardLM) and closed-source (e.g., GPT-4, PALM2, Gemini) LLMs highlights three key findings: open-source models significantly lag behind closed-source ones in format adherence; LLMs' format-following performance is independent of their content generation quality; and LLMs' format proficiency varies across different domains. These insights suggest the need for specialized tuning for format-following skills and highlight FoFo's role in guiding the selection of domain-specific AI agents. FoFo is released here at https://github.com/SalesforceAIResearch/FoFo.

Text presented in augmented reality provides in-situ, real-time information for users. However, this content can be challenging to apprehend quickly when engaging in cognitively demanding AR tasks, especially when it is presented on a head-mounted display. We propose ARTiST, an automatic text simplification system that uses a few-shot prompt and GPT-3 models to specifically optimize the text length and semantic content for augmented reality. Developed out of a formative study that included seven users and three experts, our system combines a customized error calibration model with a few-shot prompt to integrate the syntactic, lexical, elaborative, and content simplification techniques, and generate simplified AR text for head-worn displays. Results from a 16-user empirical study showed that ARTiST lightens the cognitive load and improves performance significantly over both unmodified text and text modified via traditional methods. Our work constitutes a step towards automating the optimization of batch text data for readability and performance in augmented reality.

Soybean production is susceptible to biotic and abiotic stresses, exacerbated by extreme weather events. Water limiting stress, i.e. drought, emerges as a significant risk for soybean production, underscoring the need for advancements in stress monitoring for crop breeding and production. This project combines multi-modal information to identify the most effective and efficient automated methods to investigate drought response. We investigated a set of diverse soybean accessions using multiple sensors in a time series high-throughput phenotyping manner to: (1) develop a pipeline for rapid classification of soybean drought stress symptoms, and (2) investigate methods for early detection of drought stress. We utilized high-throughput time-series phenotyping using UAVs and sensors in conjunction with machine learning (ML) analytics, which offered a swift and efficient means of phenotyping. The red-edge and green bands were most effective to classify canopy wilting stress. The Red-Edge Chlorophyll Vegetation Index (RECI) successfully differentiated susceptible and tolerant soybean accessions prior to visual symptom development. We report pre-visual detection of soybean wilting using a combination of different vegetation indices. These results can contribute to early stress detection methodologies and rapid classification of drought responses in screening nurseries for breeding and production applications.

In-context learning (ICL) exhibits dual operating modes: task learning, i.e., acquiring a new skill from in-context samples, and task retrieval, i.e., locating and activating a relevant pretrained skill. Recent theoretical work investigates various mathematical models to analyze ICL, but existing models explain only one operating mode at a time. We introduce a probabilistic model, with which one can explain the dual operating modes of ICL simultaneously. Focusing on in-context learning of linear functions, we extend existing models for pretraining data by introducing multiple task groups and task-dependent input distributions. We then analyze the behavior of the optimally pretrained model under the squared loss, i.e., the MMSE estimator of the label given in-context examples. Regarding pretraining task distribution as prior and in-context examples as the observation, we derive the closed-form expression of the task posterior distribution. With the closed-form expression, we obtain a quantitative understanding of the two operating modes of ICL. Furthermore, we shed light on an unexplained phenomenon observed in practice: under certain settings, the ICL risk initially increases and then decreases with more in-context examples. Our model offers a plausible explanation for this "early ascent" phenomenon: a limited number of in-context samples may lead to the retrieval of an incorrect skill, thereby increasing the risk, which will eventually diminish as task learning takes effect with more in-context samples. We also theoretically analyze ICL with biased labels, e.g., zero-shot ICL, where in-context examples are assigned random labels. Lastly, we validate our findings and predictions via experiments involving Transformers and large language models.

Most existing methods for predicting drug-drug interactions (DDI) predominantly concentrate on capturing the explicit relationships among drugs, overlooking the valuable implicit correlations present between drug pairs (DPs), which leads to weak predictions. To address this issue, this paper introduces a hierarchical multi-relational graph representation learning (HMGRL) approach. Within the framework of HMGRL, we leverage a wealth of drug-related heterogeneous data sources to construct heterogeneous graphs, where nodes represent drugs and edges denote clear and various associations. The relational graph convolutional network (RGCN) is employed to capture diverse explicit relationships between drugs from these heterogeneous graphs. Additionally, a multi-view differentiable spectral clustering (MVDSC) module is developed to capture multiple valuable implicit correlations between DPs. Within the MVDSC, we utilize multiple DP features to construct graphs, where nodes represent DPs and edges denote different implicit correlations. Subsequently, multiple DP representations are generated through graph cutting, each emphasizing distinct implicit correlations. The graph-cutting strategy enables our HMGRL to identify strongly connected communities of graphs, thereby reducing the fusion of irrelevant features. By combining every representation view of a DP, we create high-level DP representations for predicting DDIs. Two genuine datasets spanning three distinct tasks are adopted to gauge the efficacy of our HMGRL. Experimental outcomes unequivocally indicate that HMGRL surpasses several leading-edge methods in performance.

Automated machine learning (AutoML) aims to select and configure machine learning algorithms and combine them into machine learning pipelines tailored to a dataset at hand. For supervised learning tasks, most notably binary and multinomial classification, aka single-label classification (SLC), such AutoML approaches have shown promising results. However, the task of multi-label classification (MLC), where data points are associated with a set of class labels instead of a single class label, has received much less attention so far. In the context of multi-label classification, the data-specific selection and configuration of multi-label classifiers are challenging even for experts in the field, as it is a high-dimensional optimization problem with multi-level hierarchical dependencies. While for SLC, the space of machine learning pipelines is already huge, the size of the MLC search space outnumbers the one of SLC by several orders. In the first part of this thesis, we devise a novel AutoML approach for single-label classification tasks optimizing pipelines of machine learning algorithms, consisting of two algorithms at most. This approach is then extended first to optimize pipelines of unlimited length and eventually configure the complex hierarchical structures of multi-label classification methods. Furthermore, we investigate how well AutoML approaches that form the state of the art for single-label classification tasks scale with the increased problem complexity of AutoML for multi-label classification. In the second part, we explore how methods for SLC and MLC could be configured more flexibly to achieve better generalization performance and how to increase the efficiency of execution-based AutoML systems.

Generative artificial intelligence (GenAI) and communication networks are expected to have groundbreaking synergies in 6G. Connecting GenAI agents over a wireless network can potentially unleash the power of collective intelligence and pave the way for artificial general intelligence (AGI). However, current wireless networks are designed as a "data pipe" and are not suited to accommodate and leverage the power of GenAI. In this paper, we propose the GenAINet framework in which distributed GenAI agents communicate knowledge (high-level concepts or abstracts) to accomplish arbitrary tasks. We first provide a network architecture integrating GenAI capabilities to manage both network protocols and applications. Building on this, we investigate effective communication and reasoning problems by proposing a semantic-native GenAINet. Specifically, GenAI agents extract semantic concepts from multi-modal raw data, build a knowledgebase representing their semantic relations, which is retrieved by GenAI models for planning and reasoning. Under this paradigm, an agent can learn fast from other agents' experience for making better decisions with efficient communications. Furthermore, we conduct two case studies where in wireless device query, we show that extracting and transferring knowledge can improve query accuracy with reduced communication; and in wireless power control, we show that distributed agents can improve decisions via collaborative reasoning. Finally, we address that developing a hierarchical semantic level Telecom world model is a key path towards network of collective intelligence.