This paper describes the 5th edition of the Predicting Video Memorability Task as part of MediaEval2022. This year we have reorganised and simplified the task in order to lubricate a greater depth of inquiry. Similar to last year, two datasets are provided in order to facilitate generalisation, however, this year we have replaced the TRECVid2019 Video-to-Text dataset with the VideoMem dataset in order to remedy underlying data quality issues, and to prioritise short-term memorability prediction by elevating the Memento10k dataset as the primary dataset. Additionally, a fully fledged electroencephalography (EEG)-based prediction sub-task is introduced. In this paper, we outline the core facets of the task and its constituent sub-tasks; describing the datasets, evaluation metrics, and requirements for participant submissions.

You may not know our name but we can guarantee you know our work – all we do has a positive impact on life, health and the environment. Eurofins is by your side every day, from the food you eat to the medicines you rely on. We work with the biggest companies in the world, making sure the products they supply are safe, their ingredients are authentic and labelling is accurate. As a fast paced growing environment we are looking for natural born leaders that inspire passion in unique individuals and are not afraid to take risks in order to achieve goals. Life at Eurofins is a meritocracy, where people are empowered to make decisions and are rewarded for their success.

The Predicting Media Memorability task in the MediaEval evaluation campaign has been running annually since 2018 and several different tasks and data sets have been used in this time. This has allowed us to compare the performance of many memorability prediction techniques on the same data and in a reproducible way and to refine and improve on those techniques. The resources created to compute media memorability are now being used by researchers well beyond the actual evaluation campaign. In this paper we present a summary of the task, including the collective lessons we have learned for the research community.

Multilingual Neural Machine Translation (MNMT) models leverage many language pairs during training to improve translation quality for low-resource languages by transferring knowledge from high-resource languages. We study the quality of a domain-adapted MNMT model in the medical domain for English-Romanian with automatic metrics and a human error typology annotation which includes terminology-specific error categories. We compare the out-of-domain MNMT with the in-domain adapted MNMT. The in-domain MNMT model outperforms the out-of-domain MNMT in all measured automatic metrics and produces fewer terminology errors.

We propose a very fast frame-level model for anomaly detection in video, which learns to detect anomalies by distilling knowledge from multiple highly accurate object-level teacher models. To improve the fidelity of our student, we distill the low-resolution anomaly maps of the teachers by jointly applying standard and adversarial distillation, introducing an adversarial discriminator for each teacher to distinguish between target and generated anomaly maps. We conduct experiments on three benchmarks (Avenue, ShanghaiTech, UCSD Ped2), showing that our method is over 7 times faster than the fastest competing method, and between 28 and 62 times faster than object-centric models, while obtaining comparable results to recent methods. Our evaluation also indicates that our model achieves the best trade-off between speed and accuracy, due to its previously unheard-of speed of 1480 FPS. In addition, we carry out a comprehensive ablation study to justify our architectural design choices.

Data-driven modeling approaches such as jump tables are promising techniques to model populations of resistive random-access memory (ReRAM) or other emerging memory devices for hardware neural network simulations. As these tables rely on data interpolation, this work explores the open questions about their fidelity in relation to the stochastic device behavior they model. We study how various jump table device models impact the attained network performance estimates, a concept we define as modeling bias. Two methods of jump table device modeling, binning and Optuna-optimized binning, are explored using synthetic data with known distributions for benchmarking purposes, as well as experimental data obtained from TiOx ReRAM devices. Results on a multi-layer perceptron trained on MNIST show that device models based on binning can behave unpredictably particularly at low number of points in the device dataset, sometimes over-promising, sometimes under-promising target network accuracy. This paper also proposes device level metrics that indicate similar trends with the modeling bias metric at the network level. The proposed approach opens the possibility for future investigations into statistical device models with better performance, as well as experimentally verified modeling bias in different in-memory computing and neural network architectures.

Large language models (LLMs) transfer well to new tasks out-of-the-box simply given a natural language prompt that demonstrates how to perform the task and no additional training. Prompting is a brittle process wherein small modifications to the prompt can cause large variations in the model predictions, and therefore significant effort is dedicated towards designing a painstakingly "perfect prompt" for a task. To mitigate the high degree of effort involved in prompt-design, we instead ask whether producing multiple effective, yet imperfect, prompts and aggregating them can lead to a high quality prompting strategy. Our observations motivate our proposed prompting method, ASK ME ANYTHING (AMA). We first develop an understanding of the effective prompt formats, finding that question-answering (QA) prompts, which encourage open-ended generation ("Who went to the park?") tend to outperform those that restrict the model outputs ("John went to the park. Output True or False."). Our approach recursively uses the LLM itself to transform task inputs to the effective QA format. We apply the collected prompts to obtain several noisy votes for the input's true label. We find that the prompts can have very different accuracies and complex dependencies and thus propose to use weak supervision, a procedure for combining the noisy predictions, to produce the final predictions for the inputs. We evaluate AMA across open-source model families (e.g., EleutherAI, BLOOM, OPT, and T0) and model sizes (125M-175B parameters), demonstrating an average performance lift of 10.2% over the few-shot baseline. This simple strategy enables the open-source GPT-J-6B model to match and exceed the performance of few-shot GPT3-175B on 15 of 20 popular benchmarks. Averaged across these tasks, the GPT-J-6B model outperforms few-shot GPT3-175B. We release our code here: https://github.com/HazyResearch/ama_prompting

Most curriculum learning methods require an approach to sort the data samples by difficulty, which is often cumbersome to perform. In this work, we propose a novel curriculum learning approach termed Learning Rate Curriculum (LeRaC), which leverages the use of a different learning rate for each layer of a neural network to create a data-free curriculum during the initial training epochs. More specifically, LeRaC assigns higher learning rates to neural layers closer to the input, gradually decreasing the learning rates as the layers are placed farther away from the input. The learning rates increase at various paces during the first training iterations, until they all reach the same value. From this point on, the neural model is trained as usual. This creates a model-level curriculum learning strategy that does not require sorting the examples by difficulty and is compatible with any neural network, generating higher performance levels regardless of the architecture. We conduct comprehensive experiments on eight datasets from the computer vision (CIFAR-10, CIFAR-100, Tiny ImageNet), language (BoolQ, QNLI, RTE) and audio (ESC-50, CREMA-D) domains, considering various convolutional (ResNet-18, Wide-ResNet-50, DenseNet-121), recurrent (LSTM) and transformer (CvT, BERT, SepTr) architectures, comparing our approach with the conventional training regime. Moreover, we also compare with Curriculum by Smoothing (CBS), a state-of-the-art data-free curriculum learning approach. Unlike CBS, our performance improvements over the standard training regime are consistent across all datasets and models. Furthermore, we significantly surpass CBS in terms of training time (there is no additional cost over the standard training regime for LeRaC).

Propaganda is the expression of an opinion or an action by an individual or a group deliberately designed to influence the opinions or the actions of other individuals or groups with reference to predetermined ends, which is achieved by means of well-defined rhetorical and psychological devices. Propaganda techniques are commonly used in social media to manipulate or to mislead users. Thus, there has been a lot of recent research on automatic detection of propaganda techniques in text as well as in memes. However, so far the focus has been primarily on English. With the aim to bridge this language gap, we ran a shared task on detecting propaganda techniques in Arabic tweets as part of the WANLP 2022 workshop, which included two subtasks. Subtask~1 asks to identify the set of propaganda techniques used in a tweet, which is a multilabel classification problem, while Subtask~2 asks to detect the propaganda techniques used in a tweet together with the exact span(s) of text in which each propaganda technique appears. The task attracted 63 team registrations, and eventually 14 and 3 teams made submissions for subtask 1 and 2, respectively. Finally, 11 teams submitted system description papers.

Unstructured textual data is at the heart of healthcare systems. For obvious privacy reasons, these documents are not accessible to researchers as long as they contain personally identifiable information. One way to share this data while respecting the legislative framework (notably GDPR or HIPAA) is, within the medical structures, to de-identify it, i.e. to detect the personal information of a person through a Named Entity Recognition (NER) system and then replacing it to make it very difficult to associate the document with the person. The challenge is having reliable NER and substitution tools without compromising confidentiality and consistency in the document. Most of the conducted research focuses on English medical documents with coarse substitutions by not benefiting from advances in privacy. This paper shows how an efficient and differentially private de-identification approach can be achieved by strengthening the less robust de-identification method and by adapting state-of-the-art differentially private mechanisms for substitution purposes. The result is an approach for de-identifying clinical documents in French language, but also generalizable to other languages and whose robustness is mathematically proven.