Teacher-student knowledge distillation is a popular technique for compressing today's prevailing large language models into manageable sizes that fit low-latency downstream applications. Both the teacher and the choice of transfer set used for distillation are crucial ingredients in creating a high quality student. Yet, the generic corpora used to pretrain the teacher and the corpora associated with the downstream target domain are often significantly different, which raises a natural question: should the student be distilled over the generic corpora, so as to learn from high-quality teacher predictions, or over the downstream task corpora to align with finetuning? Our study investigates this trade-off using Domain Classification (DC) and Intent Classification/Named Entity Recognition (ICNER) as downstream tasks. We distill several multilingual students from a larger multilingual LM with varying proportions of generic and task-specific datasets, and report their performance after finetuning on DC and ICNER. We observe significant improvements across tasks and test sets when only task-specific corpora is used. We also report on how the impact of adding task-specific data to the transfer set correlates with the similarity between generic and task-specific data. Our results clearly indicate that, while distillation from a generic LM benefits downstream tasks, students learn better using target domain data even if it comes at the price of noisier teacher predictions. In other words, target domain data still trumps teacher knowledge.

Detecting Out-of-Domain (OOD) or unknown intents from user queries is essential in a task-oriented dialog system. Traditional softmax-based confidence scores are susceptible to the overconfidence issue. In this paper, we propose a simple but strong energy-based score function to detect OOD where the energy scores of OOD samples are higher than IND samples. Further, given a small set of labeled OOD samples, we introduce an energy-based margin objective for supervised OOD detection to explicitly distinguish OOD samples from INDs. Comprehensive experiments and analysis prove our method helps disentangle confidence score distributions of IND and OOD data.\footnote{Our code is available at \url{https://github.com/pris-nlp/EMNLP2022-energy_for_OOD/}.}

Understanding which student support strategies mitigate dropout and improve student retention is an important part of modern higher educational research. One of the largest challenges institutions of higher learning currently face is the scalability of student support. Part of this is due to the shortage of staff addressing the needs of students, and the subsequent referral pathways associated to provide timeous student support strategies. This is further complicated by the difficulty of these referrals, especially as students are often faced with a combination of administrative, academic, social, and socio-economic challenges. A possible solution to this problem can be a combination of student outcome predictions and applying algorithmic recommender systems within the context of higher education. While much effort and detail has gone into the expansion of explaining algorithmic decision making in this context, there is still a need to develop data collection strategies Therefore, the purpose of this paper is to outline a data collection framework specific to recommender systems within this context in order to reduce collection biases, understand student characteristics, and find an ideal way to infer optimal influences on the student journey. If confirmation biases, challenges in data sparsity and the type of information to collect from students are not addressed, it will have detrimental effects on attempts to assess and evaluate the effects of these systems within higher education.

As technology evolves, especially in the field of Artificial Intelligence, its dominance is pervading in every industry rapidly like never before and is becoming increasingly popular and valuable daily. With the use of AI, numerous autonomous applications are created to make human life and task easy. In addition, researchers are working tirelessly around the clock to make this technology more potent in the Medical Industry to provide better medical treatments, accurate diagnoses, elevate service deliveries, and more. Machine learning models are used to implement AI in the medical sector, enabling the search of healthcare data for better patient experiences and improved health results. A massive amount of machine-learning datasets and several algorithms with extraordinary decision-making capabilities are used by artificial intelligence to provide hands-on solutions based on the user requirements in every sector.

Over the last decade, Artificial intelligence (AI) has become embedded in every aspect of our society and lives. From chatbots and virtual assistants like Siri and Alexa to automated industrial machinery and self-driving cars, it's hard to ignore its impact. Today, the technology most commonly used to achieve AI is machine learning – advanced software algorithms designed to carry out one specific task, such as answering questions, translating languages or navigating a journey – and become increasingly good at it as they are exposed to more and more data. Worldwide, spending by governments and business on AI technology will top $500 billion in 2023, according to IDC research. But how will it be used, and what impact will it have?

Structured and unstructured data and facts about drugs, genes, protein, viruses, and their mechanism are spread across a huge number of scientific articles. These articles are a large-scale knowledge source and can have a huge impact on disseminating knowledge about the mechanisms of certain biological processes. A knowledge graph (KG) can be constructed by integrating such facts and data and be used for data integration, exploration, and federated queries. However, exploration and querying large-scale KGs is tedious for certain groups of users due to a lack of knowledge about underlying data assets or semantic technologies. A question-answering (QA) system allows the answer of natural language questions over KGs automatically using triples contained in a KG.

Artificial intelligence is coming to your phone and the future of AI is personal assistance. This means that AI will be able to help with your daily tasks, from making suggestions on what you should wear or buy, to answering questions about history or science. The use cases for this technology are endless but especially interesting when it comes to personal assistants and virtual assistants. In order for AI assistants to be useful, they need to make decisions that are measurable and explainable. They also have to be able to learn from their mistakes and improve over time.

"They just have to be a conservative." "I just prefer my men to be masculine." Purported conservative young women make these statements and more in an ad for new dating app "The Right Stuff," for – you guessed it – right-leaning singles. "We are living in a hyper-political environment. The biggest dealbreaker when it came to dating used to be religion, but more and more we're seeing that replaced by political affiliation," founder John McEntee said in a statement.

There has been a flurry of research in recent years on notions of fairness in ranking and recommender systems, particularly on how to evaluate if a recommender allocates exposure equally across groups of relevant items (also known as provider fairness). While this research has laid an important foundation, it gave rise to different approaches depending on whether relevant items are compared per-user/per-query or aggregated across users. Despite both being established and intuitive, we discover that these two notions can lead to opposite conclusions, a form of Simpson's Paradox. We reconcile these notions and show that the tension is due to differences in distributions of users where items are relevant, and break down the important factors of the user's recommendations. Based on this new understanding, practitioners might be interested in either notions, but might face challenges with the per-user metric due to partial observability of the relevance and user satisfaction, typical in real-world recommenders. We describe a technique based on distribution matching to estimate it in such a scenario. We demonstrate on simulated and real-world recommender data the effectiveness and usefulness of such an approach.

Movies are a great source of entertainment. However, the problem arises when one is trying to find the desired content within this vast amount of data which is significantly increasing every year. Recommender systems can provide appropriate algorithms to solve this problem. The content_based technique has found popularity due to the lack of available user data in most cases. Content_based recommender systems are based on the similarity of items' demographic information; Term Frequency _ Inverse Document Frequency (TF_IDF) and Knowledge Graph Embedding (KGE) are two approaches used to vectorize data to calculate these similarities. In this paper, we propose a weighted content_based movie RS by combining TF_IDF which is an appropriate approach for embedding textual data such as plot/description, and KGE which is used to embed named entities such as the director's name. The weights between features are determined using a Genetic algorithm. Additionally, the Iranian movies dataset is created by scraping data from movie_related websites. This dataset and the structure of the FarsBase KG are used to create the MovieFarsBase KG which is a component in the implementation process of the proposed content_based RS. Using precision, recall, and F1 score metrics, this study shows that the proposed approach outperforms the conventional approach that uses TF_IDF for embedding all attributes.