One out of four children in India are leaving grade eight without basic reading skills. Measuring the reading levels in a vast country like India poses significant hurdles. Recent advances in machine learning opens up the possibility of automating this task. However, the datasets are primarily in English. To solve this assessment problem and advance deep learning research in regional Indian languages, we present the ASER dataset of children in the age group of 6-14. The dataset consists of 5,300 subjects generating 81,658 labeled audio clips in Hindi, Marathi and English. These labels represent expert opinions on the ability of the child to read at a specified level. Using this dataset, we built a simple ASR-based classifier. Early results indicate that we can achieve a prediction accuracy of 86 percent for the English language. Considering the ASER survey spans half a million subjects, this dataset can grow to those scales.

Physics-informed neural networks (PINNs) [31] use automatic differentiation to solve partial differential equations (PDEs) by penalizing the PDE in the loss function at a random set of points in the domain of interest. Here, we develop a Petrov-Galerkin version of PINNs based on the nonlinear approximation of deep neural networks (DNNs) by selecting the {\em trial space} to be the space of neural networks and the {\em test space} to be the space of Legendre polynomials. We formulate the \textit{variational residual} of the PDE using the DNN approximation by incorporating the variational form of the problem into the loss function of the network and construct a \textit{variational physics-informed neural network} (VPINN). By integrating by parts the integrand in the variational form, we lower the order of the differential operators represented by the neural networks, hence effectively reducing the training cost in VPINNs while increasing their accuracy compared to PINNs that essentially employ delta test functions. For shallow networks with one hidden layer, we analytically obtain explicit forms of the \textit{variational residual}. We demonstrate the performance of the new formulation for several examples that show clear advantages of VPINNs over PINNs in terms of both accuracy and speed.

Current approaches to deep learning are beginning to rely heavily on transfer learning as an effective method for reducing overfitting, improving model performance, and quickly learning new tasks. Similarly, such pre-trained models are often used to create embedding representations for various types of data, such as text and images, which can then be fed as input into separate, downstream models. However, in cases where such transfer learning models perform poorly (i.e., for data outside of the training distribution), one must resort to fine-tuning such models, or even retraining them completely. Currently, no form of data augmentation has been proposed that can be applied directly to embedding inputs to improve downstream model performance. In this work, we introduce four new types of data augmentation that are generally applicable to embedding inputs, thus making them useful in both Natural Language Processing (NLP) and Computer Vision (CV) applications. For models trained on downstream tasks with such embedding inputs, these augmentation methods are shown to improve the AUC score of the models from a score of 0.9582 to 0.9812 and significantly increase the model's ability to identify classes of data that are not seen during training.

Recently the concept of transformative AI (TAI) has begun to receive attention in the AI policy space. TAI is often framed as an alternative formulation to notions of strong AI (e.g. artificial general intelligence or superintelligence) and reflects increasing consensus that advanced AI which does not fit these definitions may nonetheless have extreme and long-lasting impacts on society. However, the term TAI is poorly defined and often used ambiguously. Some use the notion of TAI to describe levels of societal transformation associated with previous 'general purpose technologies' (GPTs) such as electricity or the internal combustion engine. Others use the term to refer to more drastic levels of transformation comparable to the agricultural or industrial revolutions. The notion has also been used much more loosely, with some implying that current AI systems are already having a transformative impact on society. This paper unpacks and analyses the notion of TAI, proposing a distinction between TAI and radically transformative AI (RTAI), roughly corresponding to societal change on the level of the agricultural or industrial revolutions. We describe some relevant dimensions associated with each and discuss what kinds of advances in capabilities they might require. We further consider the relationship between TAI and RTAI and whether we should necessarily expect a period of TAI to precede the emergence of RTAI. This analysis is important as it can help guide discussions among AI policy researchers about how to allocate resources towards mitigating the most extreme impacts of AI and it can bring attention to negative TAI scenarios that are currently neglected.

Text-based games are a natural challenge domain for deep reinforcement learning algorithms. Their state and action spaces are combinatorially large, their reward function is sparse, and they are partially observable: the agent is informed of the consequences of its actions through textual feedback. In this paper we emphasize this latter point and consider the design of a deep reinforcement learning agent that can play from feedback alone. Our design recognizes and takes advantage of the structural characteristics of text-based games. We first propose a contextualisation mechanism, based on accumulated reward, which simplifies the learning problem and mitigates partial observability. We then study different methods that rely on the notion that most actions are ineffectual in any given situation, following Zahavy et al.'s idea of an admissible action. We evaluate these techniques in a series of text-based games of increasing difficulty based on the TextWorld framework, as well as the iconic game Zork. Empirically, we find that these techniques improve the performance of a baseline deep reinforcement learning agent applied to text-based games.

We present, FT-SWRL, a fuzzy temporal extension to the Semantic Web Rule Language (SWRL), which combines fuzzy theories based on the valid-time temporal model to provide a standard approach for modeling imprecise temporal domain knowledge in OWL ontologies. The proposal introduces a fuzzy temporal model for the semantic web, which is syntactically defined as a fuzzy temporal SWRL ontology (SWRL-FTO) with a new set of fuzzy temporal SWRL built-ins for defining their semantics. The SWRL-FTO hierarchically defines the necessary linguistic terminologies and variables for the fuzzy temporal model. An example model demonstrating the usefulness of the fuzzy temporal SWRL built-ins to model imprecise temporal information is also represented. Fuzzification process of interval-based temporal logic is further discussed as a reasoning paradigm for our FT-SWRL rules, with the aim of achieving a complete OWL-based fuzzy temporal reasoning. Literature review on fuzzy temporal representation approaches, both within and without the use of ontologies, led to the conclusion that the FT-SWRL model can authoritatively serve as a formal specification for handling imprecise temporal expressions on the semantic web.

In this study, we present Flatsomatic - a Variational Auto Encoder (VAE) optimized to compress somatic mutations that allow for unbiased data compression whilst maintaining the signal. We compared two different neural network architectures for the VAE: Multilayer Perceptron (MLP) and bidirectional LSTM. The somatic profiles we used to train our models consisted of 8,062 Pan-Cancer patients from The Cancer Genome Atlas and 989 cell lines from the COSMIC cell line project. The profiles for each patient were represented by the genomic loci where somatic mutations occurred and, to reduce sparsity, the locations with a frequency <5 were removed. We enhanced the VAE performance by changing its evidence lower bound, and devised an F1-score based loss showing that it helps the VAE learn better than with binary cross-entropy. We also employed beta-VAE to weight the variational regularisation term in the loss function and showed the best performance through a preliminary function to increase the weight of the regularisation term with each epoch. We assessed the reconstruction ability of the VAE using the micro F1-score metric and showed that our best performing model was a 2-layer deep MLP VAE. Our analysis also showed that the size of the latent space did not have a significant effect on the VAE learning ability. We compared the Flatsomatic embeddings created to a lower dimension version of the data from principal component analysis, showing superior performance of Flatsomatic, and performed K-means clustering on both datasets to draw comparisons to known cancer types of each profile. Finally, we present results that confirm that the Flatsomatic representations of 64 dimensions maintain the same predictive power as the original 8,298 dimensions vector, through prediction of drug response.

Source separation for music is the task of isolating contributions, or stems, from different instruments recorded individually and arranged together to form a song. Such components include voice, bass, drums and any other accompaniments. Contrarily to many audio synthesis tasks where the best performances are achieved by models that directly generate the waveform, the state-of-the-art in source separation for music is to compute masks on the magnitude spectrum. In this paper, we first show that an adaptation of Conv-Tasnet (Luo \& Mesgarani, 2019), a waveform-to-waveform model for source separation for speech, significantly beats the state-of-the-art on the MusDB dataset, the standard benchmark of multi-instrument source separation. Second, we observe that Conv-Tasnet follows a masking approach on the input signal, which has the potential drawback of removing parts of the relevant source without the capacity to reconstruct it. We propose Demucs, a new waveform-to-waveform model, which has an architecture closer to models for audio generation with more capacity on the decoder. Experiments on the MusDB dataset show that Demucs beats previously reported results in terms of signal to distortion ratio (SDR), but lower than Conv-Tasnet. Human evaluations show that Demucs has significantly higher quality (as assessed by mean opinion score) than Conv-Tasnet, but slightly more contamination from other sources, which explains the difference in SDR. Additional experiments with a larger dataset suggest that the gap in SDR between Demucs and Conv-Tasnet shrinks, showing that our approach is promising.

Local ancestry inference (LAI) allows identification of the ancestry of all chromosomal segments in admixed individuals, and it is a critical step in the analysis of human genomes with applications from pharmacogenomics and precision medicine to genome-wide association studies. In recent years, many LAI techniques have been developed in both industry and academic research. However, these methods require large training data sets of human genomic sequences from the ancestries of interest. Such reference data sets are usually limited, proprietary, protected by privacy restrictions, or otherwise not accessible to the public. Techniques to generate training samples that resemble real haploid sequences from ancestries of interest can be useful tools in such scenarios, since a generalized model can often be shared, but the unique human sample sequences cannot. In this work we present a class-conditional VAE-GAN to generate new human genomic sequences that can be used to train local ancestry inference (LAI) algorithms. We evaluate the quality of our generated data by comparing the performance of a state-of-the-art LAI method when trained with generated versus real data.

Instrument classification is one of the fields in Music Information Retrieval (MIR) that has attracted a lot of research interest. However, the majority of that is dealing with monophonic music, while efforts on polyphonic material mainly focus on predominant instrument recognition or multi-instrument recognition for entire tracks. We present an approach for instrument classification in polyphonic music using monophonic training data that involves mixing-augmentation methods. Specifically, we experiment with pitch and tempo-based synchronization, as well as mixes of tracks with similar music genres. Further, a custom CNN model is proposed, that uses the augmented training data efficiently and a plethora of suitable evaluation metrics are discussed as well. The tempo-sync and genre techniques stand out, achieving an 81% label ranking average precision accuracy, detecting up to 9 instruments in over 2300 testing tracks.