Shared benchmarks and datasets have historically played a crucial role in driving advances in large-scale applications of deep learning, e.g. in computer vision ([6]) and natural language processing ([19, 13, 11]). Neural theorem proving is a rapidly developing area which aims to apply techniques from deep learning to interactive theorem proving. To date, most contributions in this area have focused on individual theorem proving systems, each with a separately-implemented mathematics library and with results reported on a dataset-specific test split; examples include the HOList [2], CoqGym [24] and LeanStep [7] theorem proving environments and benchmarks. However, benchmarks from this paradigm are not ideal for measuring the mathematical reasoning ability of neural theorem provers for several reasons. Library-specific train/test splits are siloed by construction, dependent on how theorems and lemmas are split in these libraries, and as such are not directly comparable across systems. Moreover, formal mathematics libraries are closer to software repositories than informal mathematical exposition, and many lemmas are implementation-specific artifacts without precise informal mathematical or cross-system translations. To date, the neural theorem proving community has not organized its efforts around a cross-system benchmark. To address this need and to provide a common resource to research groups working on formal theorem proving, we present miniF2F, a unified cross-system benchmark of formal mathematics of progressively increasing difficulty, centering around Olympiad-level problem statements (AMC, AIME, IMO) as well as high-school and undergraduate maths classes.

We present an automatic non-invasive way of detecting cough events based on both accelerometer and audio signals. The acceleration signals are captured by a smartphone firmly attached to the patient's bed, using its integrated accelerometer. The audio signals are captured simultaneously by the same smartphone using an external microphone. We have compiled a manually-annotated dataset containing such simultaneously-captured acceleration and audio signals for approximately 6000 cough and 68000 non-cough events from 14 adult male patients in a tuberculosis clinic. LR, SVM and MLP are evaluated as baseline classifiers and compared with deep architectures such as CNN, LSTM, and Resnet50 using a leave-one-out cross-validation scheme. We find that the studied classifiers can use either acceleration or audio signals to distinguish between coughing and other activities including sneezing, throat-clearing, and movement on the bed with high accuracy. However, in all cases, the deep neural networks outperform the shallow classifiers by a clear margin and the Resnet50 offers the best performance by achieving an AUC exceeding 0.98 and 0.99 for acceleration and audio signals respectively. While audio-based classification consistently offers a better performance than acceleration-based classification, we observe that the difference is very small for the best systems. Since the acceleration signal requires less processing power, and since the need to record audio is sidestepped and thus privacy is inherently secured, and since the recording device is attached to the bed and not worn, an accelerometer-based highly accurate non-invasive cough detector may represent a more convenient and readily accepted method in long-term cough monitoring.

The concept of DNA storage was first suggested in 1959 by Richard Feynman who shared his vision regarding nanotechnology in the talk "There is plenty of room at the bottom". Later, towards the end of the 20-th century, the interest in storage solutions based on DNA molecules was increased as a result of the human genome project which in turn led to a significant progress in sequencing and assembly methods. DNA storage enjoys major advantages over the well-established magnetic and optical storage solutions. As opposed to magnetic solutions, DNA storage does not require electrical supply to maintain data integrity and is superior to other storage solutions in both density and durability. Given the trends in cost decreases of DNA synthesis and sequencing, it is now acknowledged that within the next 10-15 years DNA storage may become a highly competitive archiving technology and probably later the main such technology. With that said, the current implementations of DNA based storage systems are very limited and are not fully optimized to address the unique pattern of errors which characterize the synthesis and sequencing processes. In this work, we propose a robust, efficient and scalable solution to implement DNA-based storage systems. Our method deploys Deep Neural Networks (DNN) which reconstruct a sequence of letters based on imperfect cluster of copies generated by the synthesis and sequencing processes. A tailor-made Error-Correcting Code (ECC) is utilized to combat patterns of errors which occur during this process. Since our reconstruction method is adapted to imperfect clusters, our method overcomes the time bottleneck of the noisy DNA copies clustering process by allowing the use of a rapid and scalable pseudo-clustering instead. Our architecture combines between convolutions and transformers blocks and is trained using synthetic data modelled after real data statistics.

Generative adversarial networks (GANs) have an enormous potential impact on digital content creation, e.g., photo-realistic digital avatars, semantic content editing, and quality enhancement of speech and images. However, the performance of modern GANs comes together with massive amounts of computations performed during the inference and high energy consumption. That complicates, or even makes impossible, their deployment on edge devices. The problem can be reduced with quantization -- a neural network compression technique that facilitates hardware-friendly inference by replacing floating-point computations with low-bit integer ones. While quantization is well established for discriminative models, the performance of modern quantization techniques in application to GANs remains unclear. GANs generate content of a more complex structure than discriminative models, and thus quantization of GANs is significantly more challenging. To tackle this problem, we perform an extensive experimental study of state-of-art quantization techniques on three diverse GAN architectures, namely StyleGAN, Self-Attention GAN, and CycleGAN. As a result, we discovered practical recipes that allowed us to successfully quantize these models for inference with 4/8-bit weights and 8-bit activations while preserving the quality of the original full-precision models.

Mitotic figure count is an important marker of tumor proliferation and has been shown to be associated with patients' prognosis. Deep learning based mitotic figure detection methods have been utilized to automatically locate the cell in mitosis using hematoxylin \& eosin (H\&E) stained images. However, the model performance deteriorates due to the large variation of color tone and intensity in H\&E images. In this work, we proposed a two stage mitotic figure detection framework by fusing a detector and a deep ensemble classification model. To alleviate the impact of color variation in H\&E images, we utilize both stain normalization and data augmentation, aiding model to learn color irrelevant features. The proposed model obtains an F1 score of 0.7550 on the preliminary testing set released by the MIDOG challenge.

X-ray free-electron lasers (XFELs) as the world`s most brilliant light sources provide ultrashort X-ray pulses with durations typically on the order of femtoseconds. Recently, they have approached and entered the attosecond regime, which holds new promises for single-molecule imaging and studying nonlinear and ultrafast phenomena like localized electron dynamics. The technological evolution of XFELs toward well-controllable light sources for precise metrology of ultrafast processes was, however, hampered by the diagnostic capabilities for characterizing X-ray pulses at the attosecond frontier. In this regard, the spectroscopic technique of photoelectron angular streaking has successfully proven how to non-destructively retrieve the exact time-energy structure of XFEL pulses on a single-shot basis. By using artificial intelligence algorithms, in particular convolutional neural networks, we here show how this technique can be leveraged from its proof-of-principle stage toward routine diagnostics at XFELs, thus enhancing and refining their scientific access in all related disciplines.

The MS MARCO ranking dataset has been widely used for training deep learning models for IR tasks, achieving considerable effectiveness on diverse zero-shot scenarios. However, this type of resource is scarce in other languages than English. In this work we present mMARCO, a multilingual version of the MS MARCO passage ranking dataset comprising 8 languages that was created using machine translation. We evaluated mMARCO by fine-tuning mono and multilingual re-ranking models on it. Experimental results demonstrate that multilingual models fine-tuned on our translated dataset achieve superior effectiveness than models fine-tuned on the original English version alone. Also, our distilled multilingual re-ranker is competitive with non-distilled models while having 5.4 times fewer parameters. The translated datasets as well as fine-tuned models are available at https://github.com/unicamp-dl/mMARCO.git.

In this paper, we present InSeGAN, an unsupervised 3D generative adversarial network (GAN) for segmenting (nearly) identical instances of rigid objects in depth images. Using an analysis-by-synthesis approach, we design a novel GAN architecture to synthesize a multiple-instance depth image with independent control over each instance. InSeGAN takes in a set of code vectors (e.g., random noise vectors), each encoding the 3D pose of an object that is represented by a learned implicit object template. The generator has two distinct modules. The first module, the instance feature generator, uses each encoded pose to transform the implicit template into a feature map representation of each object instance. The second module, the depth image renderer, aggregates all of the single-instance feature maps output by the first module and generates a multiple-instance depth image. A discriminator distinguishes the generated multiple-instance depth images from the distribution of true depth images. To use our model for instance segmentation, we propose an instance pose encoder that learns to take in a generated depth image and reproduce the pose code vectors for all of the object instances. To evaluate our approach, we introduce a new synthetic dataset, "Insta-10", consisting of 100,000 depth images, each with 5 instances of an object from one of 10 classes. Our experiments on Insta-10, as well as on real-world noisy depth images, show that InSeGAN achieves state-of-the-art performance, often outperforming prior methods by large margins.

In this position paper, we present five key principles, namely interpretability, inherent capability to explain, independent data, interactive learning, and inquisitiveness, for the development of conversational AI that, unlike the currently popular black box approaches, is transparent and accountable. At present, there is a growing concern with the use of black box statistical language models: While displaying impressive average performance, such systems are also prone to occasional spectacular failures, for which there is no clear remedy. In an effort to initiate a discussion on possible alternatives, we outline and exemplify how our five principles enable the development of conversational AI systems that are transparent and thus safer for use. We also present some of the challenges inherent in the implementation of those principles.

Recent researches have demonstrated that BERT shows potential in a wide range of natural language processing tasks. It is adopted as an encoder for many state-of-the-art automatic summarizing systems, which achieve excellent performance. However, so far, there is not much work done for Vietnamese. In this paper, we showcase how BERT can be implemented for extractive text summarization in Vietnamese. We introduce a novel comparison between different multilingual and monolingual BERT models. The experiment results indicate that monolingual models produce promising results compared to other multilingual models and previous text summarizing models for Vietnamese.