In this paper, we propose a new variant of Linear Discriminant Analysis to overcome underlying drawbacks of traditional LDA and other LDA variants targeting problems involving imbalanced classes. Traditional LDA sets assumptions related to Gaussian class distribution and neglects influence of outlier classes, that might hurt in performance. We exploit intuitions coming from a probabilistic interpretation of visual saliency estimation in order to define saliency of a class in multi-class setting. Such information is then used to redefine the between-class and within-class scatters in a more robust manner. Compared to traditional LDA and other weight-based LDA variants, the proposed method has shown certain improvements on facial image classification problems in publicly available datasets.
The Coronavirus (COVID-19) pandemic has led to a rapidly growing `infodemic' online. Thus, the accurate retrieval of reliable relevant data from millions of documents about COVID-19 has become urgently needed for the general public as well as for other stakeholders. The COVID-19 Multilingual Information Access (MLIA) initiative is a joint effort to ameliorate exchange of COVID-19 related information by developing applications and services through research and community participation. In this work, we present a search system called Multistage BiCross Encoder, developed by team GATE for the MLIA task 2 Multilingual Semantic Search. Multistage BiCross-Encoder is a sequential three stage pipeline which uses the Okapi BM25 algorithm and a transformer based bi-encoder and cross-encoder to effectively rank the documents with respect to the query. The results of round 1 show that our models achieve state-of-the-art performance for all ranking metrics for both monolingual and bilingual runs.
Large language models have shown promising results in zero-shot settings (Brown et al.,2020; Radford et al., 2019). For example, they can perform multiple choice tasks simply by conditioning on a question and selecting the answer with the highest probability. However, ranking by string probability can be problematic due to surface form competition-wherein different surface forms compete for probability mass, even if they represent the same underlying concept, e.g. "computer" and "PC." Since probability mass is finite, this lowers the probability of the correct answer, due to competition from other strings that are valid answers (but not one of the multiple choice options). We introduce Domain Conditional Pointwise Mutual Information, an alternative scoring function that directly compensates for surface form competition by simply reweighing each option according to a term that is proportional to its a priori likelihood within the context of the specific zero-shot task. It achieves consistent gains in zero-shot performance over both calibrated (Zhao et al., 2021) and uncalibrated scoring functions on all GPT-2 and GPT-3 models over a variety of multiple choice datasets.
The article proposes a formal semantics of happiness and sadness modalities in imperfect information setting. It shows that these modalities are not definable through each other and gives a sound and complete axiomatization of their properties.
Natural Language Inference (NLI) models are known to learn from biases and artefacts within their training data, impacting how well the models generalise to other unseen datasets. While previous de-biasing approaches focus on preventing models learning from these biases, we instead provide models with information about how a human would approach the task, with the aim of encouraging the model to learn features that will generalise better to out-of-domain datasets. Using natural language explanations, we supervise a model's attention weights to encourage more attention to be paid to the words present in these explanations. For the first time, we show that training with human generated explanations can simultaneously improve performance both in-distribution and out-of-distribution for NLI, whereas most related work on robustness involves a trade-off between the two. Training with the human explanations encourages models to attend more broadly across the sentences, paying more attention to words in the premise and less attention to stop-words and punctuation. The supervised models attend to words humans believe are important, creating more robust and better performing NLI models.
Sentiment analysis of social media comments is very important for review analysis. Many online reviews are sarcastic, humorous, or hateful. This sarcastic nature of these short texts change the actual sentiments of the review as predicted by a machine learning model that attempts to detect sentiment alone. Thus, having a model that is explicitly aware of these features should help it perform better on reviews that are characterized by them. Several research has already been done in this field. This paper deals with sarcasm detection on reddit comments. Several machine learning and deep learning algorithms have been applied for the same but each of these models only take into account the initial text instead of the conversation which serves as a better measure to determine sarcasm. The other shortcoming these papers have is they rely on word embedding for representing comments and thus do not take into account the problem of polysemy(A word can have multiple meanings based on the context in which it appears). These existing modules were able to solve the problem of capturing inter sentence contextual information but not the intra sentence contextual information. So we propose a novel architecture which solves the problem of sarcasm detection by capturing intra sentence contextual information using a novel contextual attention mechanism. The proposed model solves the problem of polysemy also by using context enriched language modules like ELMO and BERT in its first component. This model comprises a total of three major components which takes into account inter sentence, intra sentence contextual information and at last use a convolutional neural network for capturing global contextual information for sarcasm detection. The proposed model was able to generate decent results and cleared showed potential to perform state of the art if trained on a larger dataset.
Sequences of events including infectious disease outbreaks, social network activities, and crimes are ubiquitous and the data on such events carry essential information about the underlying diffusion processes between communities (e.g., regions, online user groups). Modeling diffusion processes and predicting future events are crucial in many applications including epidemic control, viral marketing, and predictive policing. Hawkes processes offer a central tool for modeling the diffusion processes, in which the influence from the past events is described by the triggering kernel. However, the triggering kernel parameters, which govern how each community is influenced by the past events, are assumed to be static over time. In the real world, the diffusion processes depend not only on the influences from the past, but also the current (time-evolving) states of the communities, e.g., people's awareness of the disease and people's current interests. In this paper, we propose a novel Hawkes process model that is able to capture the underlying dynamics of community states behind the diffusion processes and predict the occurrences of events based on the dynamics. Specifically, we model the latent dynamic function that encodes these hidden dynamics by a mixture of neural networks. Then we design the triggering kernel using the latent dynamic function and its integral. The proposed method, termed DHP (Dynamic Hawkes Processes), offers a flexible way to learn complex representations of the time-evolving communities' states, while at the same time it allows to computing the exact likelihood, which makes parameter learning tractable. Extensive experiments on four real-world event datasets show that DHP outperforms five widely adopted methods for event prediction.
Graph neural networks have been widely used on modeling graph data, achieving impressive results on node classification and link prediction tasks. Yet, obtaining an accurate representation for a graph further requires a pooling function that maps a set of node representations into a compact form. A simple sum or average over all node representations considers all node features equally without consideration of their task relevance, and any structural dependencies among them. Recently proposed hierarchical graph pooling methods, on the other hand, may yield the same representation for two different graphs that are distinguished by the Weisfeiler-Lehman test, as they suboptimally preserve information from the node features. To tackle these limitations of existing graph pooling methods, we first formulate the graph pooling problem as a multiset encoding problem with auxiliary information about the graph structure, and propose a Graph Multiset Transformer (GMT) which is a multi-head attention based global pooling layer that captures the interaction between nodes according to their structural dependencies. We show that GMT satisfies both injectiveness and permutation invariance, such that it is at most as powerful as the Weisfeiler-Lehman graph isomorphism test. Moreover, our methods can be easily extended to the previous node clustering approaches for hierarchical graph pooling. Our experimental results show that GMT significantly outperforms state-of-the-art graph pooling methods on graph classification benchmarks with high memory and time efficiency, and obtains even larger performance gain on graph reconstruction and generation tasks.
Binary classifiers trained on a certain proportion of positive items introduce a bias when applied to data sets with different proportions of positive items. Most solutions for dealing with this issue assume that some information on the latter distribution is known. However, this is not always the case, certainly when this proportion is the target variable. In this paper a maximum likelihood estimator for the true proportion of positives in data sets is suggested and tested on synthetic and real world data.
A growing number of applications require the reconstructionof 3D objects from a very small number of views. In this research, we consider the problem of reconstructing a 3D object from only 4 Flash X-ray CT views taken during the impact of a Kolsky bar. For such ultra-sparse view datasets, even model-based iterative reconstruction (MBIR) methods produce poor quality results. In this paper, we present a framework based on a generalization of Plug-and-Play, known as Multi-Agent Consensus Equilibrium (MACE), for incorporating complex and nonlinear prior information into ultra-sparse CT reconstruction. The MACE method allows any number of agents to simultaneously enforce their own prior constraints on the solution. We apply our method on simulated and real data and demonstrate that MACE reduces artifacts, improves reconstructed image quality, and uncovers image features which were otherwise indiscernible.