There has been a recent explosion of computer vision models which perform many tasks and are composed of an image encoder (usually a ViT) and an autoregressive decoder (usually a Transformer). However, most of this work simply presents one system and its results, leaving many questions regarding design decisions and trade-offs of such systems unanswered. In this work, we aim to provide such answers. We take a close look at autoregressive decoders for multi-task learning in multimodal computer vision, including classification, captioning, visual question answering, and optical character recognition. Through extensive systematic experiments, we study the effects of task and data mixture, training and regularization hyperparameters, conditioning type and specificity, modality combination, and more. Importantly, we compare these to well-tuned single-task baselines to highlight the cost incurred by multi-tasking. A key finding is that a small decoder learned on top of a frozen pretrained encoder works surprisingly well. We call this setup locked-image tuning with decoder (LiT-decoder). It can be seen as teaching a decoder to interact with a pretrained vision model via natural language.

Text-to-Image (T2I) generation is enabling new applications that support creators, designers, and general end users of productivity software by generating illustrative content with high photorealism starting from a given descriptive text as a prompt. Such models are however trained on massive amounts of web data, which surfaces the peril of potential harmful biases that may leak in the generation process itself. In this paper, we take a multi-dimensional approach to studying and quantifying common social biases as reflected in the generated images, by focusing on how occupations, personality traits, and everyday situations are depicted across representations of (perceived) gender, age, race, and geographical location. Through an extensive set of both automated and human evaluation experiments we present findings for two popular T2I models: DALLE-v2 and Stable Diffusion. Our results reveal that there exist severe occupational biases of neutral prompts majorly excluding groups of people from results for both models. Such biases can get mitigated by increasing the amount of specification in the prompt itself, although the prompting mitigation will not address discrepancies in image quality or other usages of the model or its representations in other scenarios. Further, we observe personality traits being associated with only a limited set of people at the intersection of race, gender, and age. Finally, an analysis of geographical location representations on everyday situations (e.g., park, food, weddings) shows that for most situations, images generated through default location-neutral prompts are closer and more similar to images generated for locations of United States and Germany.

The popularity of generative text AI tools in answering questions has led to concerns regarding their potential negative impact on students' academic performance and the challenges that educators face in evaluating student learning. To address these concerns, this paper introduces an evolutionary approach that aims to identify the best set of Bloom's taxonomy keywords to generate questions that these tools have low confidence in answering. The effectiveness of this approach is evaluated through a case study that uses questions from a Data Structures and Representation course being taught at the University of New South Wales in Canberra, Australia. The results demonstrate that the optimization algorithm is able to find keywords from different cognitive levels to create questions that ChatGPT has low confidence in answering. This study is a step forward to offer valuable insights for educators seeking to create more effective questions that promote critical thinking among students.

The beekeeping sector has undergone considerable production variations over the past years due to adverse weather conditions, occurring more frequently as climate change progresses. These phenomena can be high-impact and cause the environment to be unfavorable to the bees' activity. We disentangle the honey production drivers with tree-based methods and predict honey production variations for hives in Italy, one of the largest honey producers in Europe. The database covers hundreds of beehive data from 2019-2022 gathered with advanced precision beekeeping techniques. We train and interpret the machine learning models making them prescriptive other than just predictive. Superior predictive performances of tree-based methods compared to standard linear techniques allow for better protection of bees' activity and assess potential losses for beekeepers for risk management.

Automated and semi-automated techniques in biomedical electron microscopy (EM) enable the acquisition of large datasets at a high rate. Segmentation methods are therefore essential to analyze and interpret these large volumes of data, which can no longer completely be labeled manually. In recent years, deep learning algorithms achieved impressive results in both pixel-level labeling (semantic segmentation) and the labeling of separate instances of the same class (instance segmentation). In this review, we examine how these algorithms were adapted to the task of segmenting cellular and sub-cellular structures in EM images. The special challenges posed by such images and the network architectures that overcame some of them are described. Moreover, a thorough overview is also provided on the notable datasets that contributed to the proliferation of deep learning in EM. Finally, an outlook of current trends and future prospects of EM segmentation is given, especially in the area of label-free learning.

Deep-sea mining would cause "extensive and irreversible" damage to sensitive habitats.NOAA This story was originally published by the Guardian and is reproduced here as part of the Climate Desk collaboration. An investigation by conservationists has found evidence that deep-seabed mining of rare minerals could cause "extensive and irreversible" damage to the planet. The report, published on Monday by the international wildlife charity Fauna & Flora, adds to the growing controversy that surrounds proposals to sweep the ocean floor of rare minerals that include cobalt, manganese and nickel. Mining companies want to exploit these deposits--which are crucial to the alternative energy sector--because land supplies are running low, they say.

But there's one thing that, as he puts it, scared the shit out of him: needles. His aversion was severe enough to hold him back from getting routine tests. Stokes, now 40, recalls an instance in his 20s when he simply couldn't bring himself to get a blood test. He once even drove to the testing facility to get his blood drawn, but couldn't follow through with it. His partner (now wife) eventually convinced him to get the test, but he remembers it as one of "the most horrific" experiences he's had. "I kind of passed out a little bit along the way, and was sweaty and clammy and all that sort of stuff," he said.

Founded in Sydney in 2017, we saw the need to provide short term, flexible leases of electric bikes to people who want to make money from food delivery. We are doubling in size every 6 months! The problems you solve and the impact you have will be key to helping Zoomo lead the transformation of last-mile logistics as we scale across global markets. We are transitioning billions of urban delivery miles from bikes, cars and trucks to smart e-bikes, starting with food delivery. At Zoomo, we celebrate diversity and are committed to creating an inclusive environment and equal opportunities.

Evaluating the robustness of a defense model is a challenging task in adversarial robustness research. Obfuscated gradients have previously been found to exist in many defense methods and cause a false signal of robustness. In this paper, we identify a more subtle situation called Imbalanced Gradients that can also cause overestimated adversarial robustness. The phenomenon of imbalanced gradients occurs when the gradient of one term of the margin loss dominates and pushes the attack towards to a suboptimal direction. To exploit imbalanced gradients, we formulate a Margin Decomposition (MD) attack that decomposes a margin loss into individual terms and then explores the attackability of these terms separately via a two-stage process. We also propose a multi-targeted and ensemble version of our MD attack. By investigating 24 defense models proposed since 2018, we find that 11 models are susceptible to a certain degree of imbalanced gradients and our MD attack can decrease their robustness evaluated by the best standalone baseline attack by more than 1%. We also provide an in-depth investigation on the likely causes of imbalanced gradients and effective countermeasures.

We define `ousiometrics' to be the study of essential meaning in whatever context that meaningful signals are communicated, and `telegnomics' as the study of remotely sensed knowledge. From work emerging through the middle of the 20th century, the essence of meaning has become generally accepted as being well captured by the three orthogonal dimensions of evaluation, potency, and activation (EPA). By re-examining first types and then tokens for the English language, and through the use of automatically annotated histograms -- `ousiograms' -- we find here that: 1. The essence of meaning conveyed by words is instead best described by a compass-like power-danger (PD) framework, and 2. Analysis of a disparate collection of large-scale English language corpora -- literature, news, Wikipedia, talk radio, and social media -- shows that natural language exhibits a systematic bias toward safe, low danger words -- a reinterpretation of the Pollyanna principle's positivity bias for written expression. To help justify our choice of dimension names and to help address the problems with representing observed ousiometric dimensions by bipolar adjective pairs, we introduce and explore `synousionyms' and `antousionyms' -- ousiometric counterparts of synonyms and antonyms. We further show that the PD framework revises the circumplex model of affect as a more general model of state of mind. Finally, we use our findings to construct and test a prototype `ousiometer', a telegnomic instrument that measures ousiometric time series for temporal corpora. We contend that our power-danger ousiometric framework provides a complement for entropy-based measurements, and may be of value for the study of a wide variety of communication across biological and artificial life.