Language models are defined over a finite set of inputs, which creates a vocabulary bottleneck when we attempt to scale the number of supported languages. Tackling this bottleneck results in a trade-off between what can be represented in the embedding matrix and computational issues in the output layer. This paper introduces PIXEL, the Pixel-based Encoder of Language, which suffers from neither of these issues. PIXEL is a pretrained language model that renders text as images, making it possible to transfer representations across languages based on orthographic similarity or the co-activation of pixels. PIXEL is trained to reconstruct the pixels of masked patches instead of predicting a distribution over tokens. We pretrain the 86M parameter PIXEL model on the same English data as BERT and evaluate on syntactic and semantic tasks in typologically diverse languages, including various non-Latin scripts. We find that PIXEL substantially outperforms BERT on syntactic and semantic processing tasks on scripts that are not found in the pretraining data, but PIXEL is slightly weaker than BERT when working with Latin scripts. Furthermore, we find that PIXEL is more robust than BERT to orthographic attacks and linguistic code-switching, further confirming the benefits of modelling language with pixels.

Labor shortages in fruit crop production have prompted the development of mechanized and automated machines as alternatives to labor-intensive orchard operations such as harvesting, pruning, and thinning. Agricultural robots capable of identifying tree canopy parts and estimating geometric and topological parameters, such as branch diameter, length, and angles, can optimize crop yields through automated pruning and thinning platforms. In this study, we proposed a machine vision system to estimate canopy parameters in apple orchards and determine an optimal number of fruit for individual branches, providing a foundation for robotic pruning, flower thinning, and fruitlet thinning to achieve desired yield and quality.Using color and depth information from an RGB-D sensor (Microsoft Azure Kinect DK), a YOLOv8-based instance segmentation technique was developed to identify trunks and branches of apple trees during the dormant season. Principal Component Analysis was applied to estimate branch diameter (used to calculate limb cross-sectional area, or LCSA) and orientation. The estimated branch diameter was utilized to calculate LCSA, which served as an input for crop-load estimation, with larger LCSA values indicating a higher potential fruit-bearing capacity.RMSE for branch diameter estimation was 2.08 mm, and for crop-load estimation, 3.95. Based on commercial apple orchard management practices, the target crop-load (number of fruit) for each segmented branch was estimated with a mean absolute error (MAE) of 2.99 (ground truth crop-load was 6 apples per LCSA). This study demonstrated a promising workflow with high performance in identifying trunks and branches of apple trees in dynamic commercial orchard environments and integrating farm management practices into automated decision-making.

Publicly available information contains valuable information for Cyber Threat Intelligence (CTI). This can be used to prevent attacks that have already taken place on other systems. Ideally, only the initial attack succeeds and all subsequent ones are detected and stopped. But while there are different standards to exchange this information, a lot of it is shared in articles or blog posts in non-standardized ways. Manually scanning through multiple online portals and news pages to discover new threats and extracting them is a time-consuming task. To automize parts of this scanning process, multiple papers propose extractors that use Natural Language Processing (NLP) to extract Indicators of Compromise (IOCs) from documents. However, while this already solves the problem of extracting the information out of documents, the search for these documents is rarely considered. In this paper, a new focused crawler is proposed called ThreatCrawl, which uses Bidirectional Encoder Representations from Transformers (BERT)-based models to classify documents and adapt its crawling path dynamically. While ThreatCrawl has difficulties to classify the specific type of Open Source Intelligence (OSINT) named in texts, e.g., IOC content, it can successfully find relevant documents and modify its path accordingly. It yields harvest rates of up to 52%, which are, to the best of our knowledge, better than the current state of the art.

Constrained multiagent reinforcement learning (C-MARL) is gaining importance as MARL algorithms find new applications in real-world systems ranging from energy systems to drone swarms. Most C-MARL algorithms use a primal-dual approach to enforce constraints through a penalty function added to the reward. In this paper, we study the structural effects of this penalty term on the MARL problem. First, we show that the standard practice of using the constraint function as the penalty leads to a weak notion of safety. However, by making simple modifications to the penalty term, we can enforce meaningful probabilistic (chance and conditional value at risk) constraints. Second, we quantify the effect of the penalty term on the value function, uncovering an improved value estimation procedure. We use these insights to propose a constrained multiagent advantage actor critic (C-MAA2C) algorithm. Simulations in a simple constrained multiagent environment affirm that our reinterpretation of the primal-dual method in terms of probabilistic constraints is effective, and that our proposed value estimate accelerates convergence to a safe joint policy.

From health to education, income impacts a huge range of life choices. Earlier research has leveraged data from online social networks to study precisely this impact. In this paper, we ask the opposite question: do different levels of income result in different online behaviors? We demonstrate it does. We present the first large-scale study of Nextdoor, a popular location-based social network. We collect 2.6 Million posts from 64,283 neighborhoods in the United States and 3,325 neighborhoods in the United Kingdom, to examine whether online discourse reflects the income and income inequality of a neighborhood. We show that posts from neighborhoods with different incomes indeed differ, e.g. richer neighborhoods have a more positive sentiment and discuss crimes more, even though their actual crime rates are much lower. We then show that user-generated content can predict both income and inequality. We train multiple machine learning models and predict both income (R-squared=0.841) and inequality (R-squared=0.77).

Models of actual causality leverage domain knowledge to generate convincing diagnoses of events that caused an outcome. It is promising to apply these models to diagnose and repair run-time property violations in cyber-physical systems (CPS) with learning-enabled components (LEC). However, given the high diversity and complexity of LECs, it is challenging to encode domain knowledge (e.g., the CPS dynamics) in a scalable actual causality model that could generate useful repair suggestions. In this paper, we focus causal diagnosis on the input/output behaviors of LECs. Specifically, we aim to identify which subset of I/O behaviors of the LEC is an actual cause for a property violation. An important by-product is a counterfactual version of the LEC that repairs the run-time property by fixing the identified problematic behaviors. Based on this insights, we design a two-step diagnostic pipeline: (1) construct and Halpern-Pearl causality model that reflects the dependency of property outcome on the component's I/O behaviors, and (2) perform a search for an actual cause and corresponding repair on the model. We prove that our pipeline has the following guarantee: if an actual cause is found, the system is guaranteed to be repaired; otherwise, we have high probabilistic confidence that the LEC under analysis did not cause the property violation. We demonstrate that our approach successfully repairs learned controllers on a standard OpenAI Gym benchmark.

Text-conditioned image generation models have recently achieved astonishing results in image quality and text alignment and are consequently employed in a fast-growing number of applications. Since they are highly data-driven, relying on billion-sized datasets randomly scraped from the internet, they also suffer, as we demonstrate, from degenerated and biased human behavior. In turn, they may even reinforce such biases. To help combat these undesired side effects, we present safe latent diffusion (SLD). Specifically, to measure the inappropriate degeneration due to unfiltered and imbalanced training sets, we establish a novel image generation test bed-inappropriate image prompts (I2P)-containing dedicated, real-world image-to-text prompts covering concepts such as nudity and violence. As our exhaustive empirical evaluation demonstrates, the introduced SLD removes and suppresses inappropriate image parts during the diffusion process, with no additional training required and no adverse effect on overall image quality or text alignment.

Ensuring the realism of computer-generated synthetic images is crucial to deep neural network (DNN) training. Due to different semantic distributions between synthetic and real-world captured datasets, there exists semantic mismatch between synthetic and refined images, which in turn results in the semantic distortion. Recently, contrastive learning (CL) has been successfully used to pull correlated patches together and push uncorrelated ones apart. In this work, we exploit semantic and structural consistency between synthetic and refined images and adopt CL to reduce the semantic distortion. Besides, we incorporate hard negative mining to improve the performance furthermore. We compare the performance of our method with several other benchmarking methods using qualitative and quantitative measures and show that our method offers the state-of-the-art performance.

Artificial intelligence, machine learning, and algorithmic techniques in general, provide two crucial abilities with the potential to improve decision-making in the context of allocation of scarce societal resources. They have the ability to flexibly and accurately model treatment response at the individual level, potentially allowing us to better match available resources to individuals. In addition, they have the ability to reason simultaneously about the effects of matching sets of scarce resources to populations of individuals. In this work, we leverage these abilities to study algorithmic allocation of scarce societal resources in the context of homelessness. In communities throughout the United States, there is constant demand for an array of homeless services intended to address different levels of need. Allocations of housing services must match households to appropriate services that continuously fluctuate in availability, while inefficiencies in allocation could “waste” scarce resources as households will remain in-need and re-enter the homeless system, increasing the overall demand for homeless services. This complex allocation problem introduces novel technical and ethical challenges. Using administrative data from a regional homeless system, we formulate the problem of “optimal” allocation of resources given data on households with need for homeless services. The optimization problem aims to allocate available resources such that predicted probabilities of household re-entry are minimized. The key element of this work is its use of a counterfactual prediction approach that predicts household probabilities of re-entry into homeless services if assigned to each service. Through these counterfactual predictions, we find that this approach has the potential to improve the efficiency of the homeless system by reducing re-entry, and, therefore, system-wide demand. However, efficiency comes with trade-offs - a significant fraction of households are assigned to services that increase probability of re-entry. To address this issue as well as the inherent fairness considerations present in any context where there are insufficient resources to meet demand, we discuss the efficiency, equity, and fairness issues that arise in our work and consider potential implications for homeless policies.

Organizations are increasingly adopting machine learning (ML) for personnel assessment. However, concerns exist about fairness in designing and implementing ML assessments. Supervised ML models are trained to model patterns in data, meaning ML models tend to yield predictions that reflect subgroup differences in applicant attributes in the training data, regardless of the underlying cause of subgroup differences. In this study, we systematically under- and oversampled minority (Black and Hispanic) applicants to manipulate adverse impact ratios in training data and investigated how training data adverse impact ratios affect ML model adverse impact and accuracy. We used self-reports and interview transcripts from job applicants (N = 2,501) to train 9,702 ML models to predict screening decisions. We found that training data adverse impact related linearly to ML model adverse impact. However, removing adverse impact from training data only slightly reduced ML model adverse impact and tended to negatively affect ML model accuracy. We observed consistent effects across self-reports and interview transcripts, whether oversampling real (i.e., bootstrapping) or synthetic observations. As our study relied on limited predictor sets from one organization, the observed effects on adverse impact may be attenuated among more accurate ML models.