The current methodology in tackling Acoustic Scene Classification (ASC) task can be described in two steps, preprocessing of the audio waveform into log-mel spectrogram and then using it as the input representation for Convolutional Neural Network (CNN). This paradigm shift occurs after DCASE 2016 where this framework model achieves the state-of-the-art result in ASC tasks on the (ESC-50) dataset and achieved an accuracy of 64.5%, which constitute to 20.5% improvement over the baseline model, and DCASE 2016 dataset with an accuracy of 90.0% (development) and 86.2% (evaluation), which constitute a 6.4% and 9% improvements with respect to the baseline system. In this paper, we explored the use of harmonic and percussive source separation (HPSS) to split the audio into harmonic audio and percussive audio, which has received popularity in the field of music information retrieval (MIR). Although works have been done in using HPSS as input representation for CNN model in ASC task, this paper further investigate the possibility on leveraging the separated harmonic component and percussive component by curating 2 CNNs which tries to understand harmonic audio and percussive audio in their natural form, one specialized in extracting deep features in time biased domain and another specialized in extracting deep features in frequency biased domain, respectively. The deep features extracted from these 2 CNNs will then be combined using bilinear pooling. Hence, presenting a two-stream time and frequency CNN architecture approach in classifying acoustic scene. The model is being evaluated on DCASE 2019 sub task 1a dataset and scored an average of 65% on development dataset, Kaggle Leadership Private and Public board.

Alphabet's Jigsaw subsidiary has revealed a new tool that's intended to help journalists more easily spot deepfakes and manipulated images. The tool, called Assembler, uses "detectors" to analyze an image or deepfake; determine if it's authentic; and if it's not, inform the user where the image may have been manipulated. Assembler's detectors can see things like an image pasted over another or changes to image brightness, according to a blog post by Jigsaw CEO and founder Jared Cohen. Jigsaw also says Assembler has a detector specifically made to catch deepfakes and one that takes the combined signals from many detectors to analyze an image for multiple instances of manipulation at the same time. Right now, the tool is in testing with more than a dozen news and fact-checking organizations, including Animal Politico, Rappler, and Agence France-Presse, though Jigsaw doesn't plan to offer it to the public, according to The New York Times.

EVENT: The UCI Forum for the Academy and the Public will host a two-day conference on "The Future of the Future: The Ethics and Implications of AI." Keynote speaker Bruce Sterling, an award-winning science fiction author, and an interdisciplinary and international panel of writers, academics and communicators will tackle an advance in technology that touches our everyday lives: artificial intelligence. INFORMATION: All events are free and open to the public, but please RSVP here. Visitor parking is available in the Student Center Parking Structure (grid D5 on campus map) and in the Mesa Parking Structure (grid D3 on campus map) for $13 per day or $2 per hour. Media planning to attend should contact Pat Harriman at 949-824-9055 or pharrima@uci.edu. BACKGROUND: The UCI Forum for the Academy and the Public is a collaborative project of the literary journalism program, the School of Humanities and the School of Law that bridges the university and the public via conferences and pop-ups that take on the most pressing issues of our time.

As we look to the future to TV advertising, one thing is certain: Broadcast networks have had a difficult time fighting the powers of choice and it is this choice that has shaped a new television landscape where audiences can now enjoy a plethora of platforms. Competition for viewers has challenged traditional content owners and distributors as well as traditional advertisers, as all have experienced an erosion of audience. Broadcasters have had to "guarantee" ratings to keep their advertising rates, relying on a forecasted rating to demonstrate viewership of a particular show. Most efforts in forecasting, however, have not been reliable, let alone validated. From the simple empirical model, to regression models, to the even more sophisticated Bayesian model-averaging methods, several ratings forecasting providers have come and gone.

Audio features have been proven useful for increasing the performance of automated topic segmentation systems. This study explores the novel task of using audio embeddings for automated, topically coherent segmentation of radio shows. We created three different audio embedding generators using multi-class classification tasks on three datasets from different domains. We evaluate topic segmentation performance of the audio embeddings and compare it against a text-only baseline. We find that a set-up including audio embeddings generated through a non-speech sound event classification task significantly outperforms our text-only baseline by 32.3% in F1-measure. In addition, we find that different classification tasks yield audio embeddings that vary in segmentation performance.

Recent work has shown great promise in explaining neural network behavior. In particular, feature attribution methods explain which features were most important to a model's prediction on a given input. However, for many tasks, simply knowing which features were important to a model's prediction may not provide enough insight to understand model behavior. The interactions between features within the model may better help us understand not only the model, but also why certain features are more important than others. In this work we present Integrated Hessians, an extension of Integrated Gradients that explains pairwise feature interactions in neural networks. Integrated Hessians overcomes several theoretical limitations of previous methods to explain interactions, and unlike such previous methods is not limited to a specific architecture or class of neural network. We apply Integrated Hessians on a variety of neural networks trained on language data, biological data, astronomy data, and medical data and gain new insight into model behavior in each domain. Code available at https://github.com/suinleelab/path_explain

Active speaker detection (ASD) and virtual cinematography (VC) can significantly improve the remote user experience of a video conference by automatically panning, tilting and zooming of a video conferencing camera: users subjectively rate an expert video cinematographer's video significantly higher than unedited video. We describe a new automated ASD and VC that performs within 0.3 MOS of an expert cinematographer based on subjective ratings with a 1-5 scale. This system uses a 4K wide-FOV camera, a depth camera, and a microphone array; it extracts features from each modality and trains an ASD using an AdaBoost machine learning system that is very efficient and runs in real-time. A VC is similarly trained using machine learning to optimize the subjective quality of the overall experience. To avoid distracting the room participants and reduce switching latency the system has no moving parts -- the VC works by cropping and zooming the 4K wide-FOV video stream. The system was tuned and evaluated using extensive crowdsourcing techniques and evaluated on a dataset with N=100 meetings, each 2-5 minutes in length.

Fox News Flash top entertainment and celebrity headlines are here. Check out what's clicking today in entertainment. Is Ben Affleck on a dating app or not? According to Patti Stanger, he is. The "Millionaire Matchmaker" star recently appeared on Us Weekly's "Hot Hollywood" podcast, where she claimed she's seen the 47-year-old actor on the elite dating app, Raya, on multiple occasions.

For the next few weeks of AI Show, we are taking a bit of a different tack (let us know if you want more or less of this kind of content).

Traditional machine learning methods that are used to detect threats at the machine layer aren't equipped to account for the complexities of human …