Mathematical Information Retrieval (MIR) has been actively studied in recent years and many fruitful results have emerged. Among those, the Approach Zero system is one of the few math-aware search engines that is able to perform substructure matching efficiently. Furthermore, it has been deployed in ARQMath2020, the most recent community-wide MIR evaluation, as a strong baseline due to its empirical effectiveness and ability to handle structured math content. However, in order to implement a retrieval model that handles structured queries efficiently, Approach Zero is written in C from the ground up, requiring special pipelines for processing math content and queries. Thus, the system is not conveniently accessible and reusable to the community as a research tool. In this paper, we present PyA0, an easy-to-use Python toolkit built on Approach Zero that improves its accessibility to researchers. We introduce the toolkit interface and report evaluation results on popular MIR datasets to demonstrate the effectiveness and efficiency of our toolkit. We have made PyA0 source code publicly accessible at https://github.com/approach0/pya0, which includes a link to a notebook demo.
We consider a source sensed by several correlated sensors. Each one of the sensors receives a noisy version of the source, performs channel coding, and transmits this information through an independent noisy channel to a central processing unit. The processing unit fuses all the received sequences with the objective of obtaining the best possible estimate of the source of interest. This is performed by utilizing at the decoder a graphical model that jointly describes the channel decoders for each sensor and the probabilistic dependence between the source and the sensors.
We investigate the use of LDGM codes for the transmission of two highly correlated senders over a Rayleigh fading multiple access channel. Each one of the senders is encoded independently, and without knowledge of the correlation parameters, using an LDGM code. Decoding is performed in a common receiver, which exploits the correlation between the senders by performing message passing in a graph jointly describing the correlation model and the channel codes. The resulting performance surpasses the theoretical limits obtained when separation between source and channel coding is assumed. As opposed to the case of uncorrelated senders, the performance loss when the channel state information is not available at the decoder is practically non-existent.
With the recent success of dense retrieval methods based on bi-encoders, studies have applied this approach to various interesting downstream retrieval tasks with good efficiency and in-domain effectiveness. Recently, we have also seen the presence of dense retrieval models in Math Information Retrieval (MIR) tasks, but the most effective systems remain classic retrieval methods that consider hand-crafted structure features. In this work, we try to combine the best of both worlds:\ a well-defined structure search method for effective formula search and efficient bi-encoder dense retrieval models to capture contextual similarities. Specifically, we have evaluated two representative bi-encoder models for token-level and passage-level dense retrieval on recent MIR tasks. Our results show that bi-encoder models are highly complementary to existing structure search methods, and we are able to advance the state-of-the-art on MIR datasets.