Recent advances in deep learning have expanded the application of large language models (LLMs) across fields such as medicine, finance, and education. Understanding the mechanisms underlying these models is essential to mitigate issues like hallucinations and bias. This study provides deep learning practitioners with insights into how specific training data points and internal structures influence model behaviour. Using influence functions and mechanistic interpretability, we will analyze the impact of data on model predictions across various tasks. Preliminary findings indicate that semantic search techniques, such as FAISS, enable efficient identification of influential training points in GPT-2 small. Future work will extend these methods to additional tasks and more complex models, with a focus on further elucidating LLM structures to improve interpretability.

Most of the previous neural machine translations (NMT) rely on parallel corpus. Integrating explicitly prior syntactic structure information can improve the neural machine translation. In this article, we propose a Syntax Induced Self-Attention (SISA) which explores the influence of dependence relation between words through the attention mechanism and fine-tunes the attention allocation of the sentence through the obtained dependency weight. We present a new model, Double Syntax Induced Self-Attention (DSISA), which fuses the features extracted by SISA and a compact convolution neural network (CNN). SISA can alleviate long dependency in sentence, while CNN captures the limited context based on neighbors. DSISA utilizes two different neural networks to extract different features for richer semantic representation and replaces the first layer of Transformer encoder. DSISA not only makes use of the global feature of tokens in sentences but also the local feature formed with adjacent tokens. Finally, we perform simulation experiments that verify the performance of the new model on standard corpora.

Various innovative and original works have been applied and proposed in the field of sports video analysis. However, individual works have focused on sophisticated methodologies with particular sport types and there has been a lack of scalable and holistic frameworks in this field. This article proposes a solution and presents a systematic and generic approach which is experimented on a relatively large-scale sports consortia. The system aims at the event detection scenario of an input video with an orderly sequential process. Initially, domain knowledge-independent local descriptors are extracted homogeneously from the input video sequence. Then the video representation is created by adopting a bag-of-visual-words (BoW) model. The video’s genre is first identified by applying the k-nearest neighbor (k-NN) classifiers on the initially obtained video representation, and various dissimilarity measures are assessed and evaluated analytically. Subsequently, an unsupervised probabilistic latent semantic analysis (PLSA)-based approach is employed at the same histogram-based video representation, characterizing each frame of video sequence into one of four view groups, namely closed-up-view, mid-view, long-view, and outer-field-view. Finally, a hidden conditional random field (HCRF) structured prediction model is utilized for interesting event detection. From experimental results, k-NN classifier using KL-divergence measurement demonstrates the best accuracy at 82.16% for genre categorization. Supervised SVM and unsupervised PLSA have average classification accuracies at 82.86% and 68.13%, respectively. The HCRF model achieves 92.31% accuracy using the unsupervised PLSA based label input, which is comparable with the supervised SVM based input at an accuracy of 93.08%. In general, such a systematic approach can be widely applied in processing massive videos generically.

This paper presents a framework with two automatic tasks targeting large-scale and low quality sports video archives collected from online video streams. The framework is based on the bag of visual-words model using speeded-up robust features (SURF). The first task is sports genre categorization based on hierarchical structure. Following on the second task which is based on automatically obtained genre, views are classified using support vector machines (SVMs). As a consequence, the views classification result can be used in video parsing and highlight extraction. As compared with state-of-the-art methods, our approach is fully automatic as well as domain knowledge free and thus provides a better extensibility. Furthermore, our dataset consists of 14 sport genres with 6850 minutes in total. Both sport genre categorization and view type classification have more than 80% accuracy rates, which validate this framework's robustness and potential in web-based applications.