This paper addresses the problem of training a deep neural network for satellite image segmentation so that it can be deployed over images whose statistics differ from those used for training. For example, in postdisaster damage assessment, the tight time constraints make it impractical to train a network from scratch for each image to be segmented. We propose a convolutional encoder-decoder network able to learn visual representations of increasing semantic level as its depth increases, allowing it to generalize over a wider range of satellite images. Then, we propose two additional methods to improve the network performance over each specific image to be segmented. First, we observe that updating the batch normalization layers' statistics over the target image improves the network performance without human intervention. Second, we show that refining a trained network over a few samples of the image boosts the network performance with minimal human intervention. We evaluate our architecture over three data sets of satellite images, showing the state-of-the-art performance in binary segmentation of previously unseen images and competitive performance with respect to more complex techniques in a multiclass segmentation task.

The advent of Artificial Intelligence (AI) technologies has precipitated the rise of asynchronous video interviews (AVIs) as an alternative to conventional job interviews. These one-way video interviews are conducted online and can be analyzed using AI algorithms to automate and speed up the selection procedure. In particular, the swift advancement of Large Language Models (LLMs) has significantly decreased the cost and technical barrier to developing AI systems for automatic personality and interview performance evaluation. However, the generative and task-unspecific nature of LLMs might pose potential risks and biases when evaluating humans based on their AVI responses. In this study, we conducted a comprehensive evaluation of the validity, reliability, fairness, and rating patterns of two widely-used LLMs, GPT-3.5 and GPT-4, in assessing personality and interview performance from an AVI. We compared the personality and interview performance ratings of the LLMs with the ratings from a task-specific AI model and human annotators using simulated AVI responses of 685 participants. The results show that LLMs can achieve similar or even better zero-shot validity compared with the task-specific AI model when predicting personality traits. The verbal explanations for predicting personality traits generated by LLMs are interpretable by the personality items that are designed according to psychological theories. However, LLMs also suffered from uneven performance across different traits, insufficient test-retest reliability, and the emergence of certain biases. Thus, it is necessary to exercise caution when applying LLMs for human-related application scenarios, especially for significant decisions such as employment.

In two studies, we examined the effects of algorithm‐based (vs. recruiter‐based) evaluations of an asynchronous video interview and a personality inventory on applicant reactions. In line with our expectations, we found several negative applicant reactions to the use of algorithms. Specifically, in Study 1 ( N = 172), informing participants that an algorithm, rather than a recruiter, had analysed their interview and personality inventory increased feelings of emotional creepiness, and reduced fairness perceptions, perceived predictive validity and feedback acceptance. In Study 2 ( N = 276), we were able to replicate these effects for fairness perceptions and perceived predictive validity. Furthermore, in both studies, algorithm‐based evaluations negatively affected feedback acceptance, organizational attraction and job acceptance intentions through fairness perceptions. However, in contrast with our expectations, selection decision favourability did not influence the impact of evaluation source (recruiter vs. algorithm) on applicant reactions. In Study 2, we also found some tentative evidence that applicant reactions to algorithm‐based evaluations are not affected by the type of information source (i.e. verbal vs. nonverbal cues) on which the algorithm is based.

Recent advances in AI-based learning models have significantly increased the accuracy of Automatic Personality Recognition (APR). However, these methods either require training data from the same subject or the meta-information from the training set to learn the personality-related features (i.e., subject-dependency). The variance of feature extraction for different subjects compromises the possibility of designing a dependency-free system for APR. To address this problem, we present an unsupervised multimodal learning framework to infer personality traits from audio, visual, and verbal modalities. Our method both extracts the handcraft features and transfers deep-learning based embeddings from other tasks (e.g., emotion recognition) to recognize personality traits. Since these representations are extracted locally in the time domain, we present an unsupervised temporal aggregation method to aggregate the extracted features over the temporal dimension. We evaluate our method on the ChaLearn dataset, the most widely referenced dataset for APR, using a dependency-free split of the dataset. Our results show that the proposed feature extraction and temporal aggregation modules do not require personality annotations in training but still outperform other state-of-the-art baseline methods. We also address the problem of subject-dependency in the original split of the ChaLearn dataset. The newly proposed split (i.e., data for training, validation, and testing) of the dataset can benefit the community by providing a more accurate method to validate the subject-generalizability of APR algorithms.

A cloud screening unit on a satellite platform for Earth observation can play an important role in optimizing communication resources by selecting images with interesting content while skipping those that are highly contaminated by clouds. In this study, we address the cloud screening problem by investigating an encoder–decoder convolutional neural network (CNN). CNNs usually employ millions of parameters to provide high accuracy; on the other hand, the satellite platform imposes hardware constraints on the processing unit. Hence, to allow an onboard implementation, we investigate experimentally several solutions to reduce the resource consumption by CNN while preserving its classification accuracy. We experimentally explore approaches such as halving the computation precision, using fewer spectral bands, reducing the input size, decreasing the number of network filters and also making use of shallower networks, with the constraint that the resulting CNN must have sufficiently small memory footprint to fit the memory of a low-power accelerator for embedded systems. The trade-off between the network performance and resource consumption has been studied over the publicly available SPARCS dataset. Finally, we show that the proposed network can be implemented on the satellite board while performing with reasonably high accuracy compared with the state-of-the-art.