The increasing capabilities of large generative models and their ever more widespread deployment have raised concerns about their reliability, safety, and potential misuse. To address these issues, recent works have proposed to control model generation by steering model activations in order to effectively induce or prevent the emergence of concepts or behaviors in the generated output. In this paper we introduce Activation Transport (AcT), a general framework to steer activations guided by optimal transport theory that generalizes many previous activation-steering works. AcT is modality-agnostic and provides fine-grained control over the model behavior with negligible computational overhead, while minimally impacting model abilities. We experimentally show the effectiveness and versatility of our approach by addressing key challenges in large language models (LLMs) and text-to-image diffusion models (T2Is). For LLMs, we show that AcT can effectively mitigate toxicity, induce arbitrary concepts, and increase their truthfulness. In T2Is, we show how AcT enables fine-grained style control and concept negation.

The recent rapid adoption of large language models (LLMs) highlights the critical need for benchmarking their fairness. Conventional fairness metrics, which focus on discrete accuracy-based evaluations (i.e., prediction correctness), fail to capture the implicit impact of model uncertainty (e.g., higher model confidence about one group over another despite similar accuracy). To address this limitation, we propose an uncertainty-aware fairness metric, UCerf, to enable a fine-grained evaluation of model fairness that is more reflective of the internal bias in model decisions. Furthermore, observing data size, diversity, and clarity issues in current datasets, we introduce a new gender-occupation fairness evaluation dataset with 31,756 samples for co-reference resolution, offering a more diverse and suitable benchmark for modern LLMs. Combining our metric and dataset, we provide insightful comparisons of eight open-source LLMs. For example, Mistral-8B exhibits suboptimal fairness due to high confidence in incorrect predictions, a detail overlooked by Equalized Odds but captured by UCerF. Overall, this work provides a holistic framework for LLM evaluation by jointly assessing fairness and uncertainty, enabling the development of more transparent and accountable AI systems.

An important issue with Large Language Models (LLMs) is their undesired ability to generate toxic language. In this work, we show that the neurons responsible for toxicity can be determined by their power to discriminate toxic sentences, and that toxic language can be mitigated by reducing their activation levels proportionally to this power. We propose AUROC adaptation (AurA), an intervention that can be applied to any pre-trained LLM to mitigate toxicity. As the intervention is proportional to the ability of each neuron to discriminate toxic content, it is free of any model-dependent hyperparameters. We show that AurA can achieve up to $2.2\times$ reduction in toxicity with only a $0.72$ perplexity increase. We also show that AurA is effective with models of different scale (from 1.5B to 40B parameters), and its effectiveness in mitigating toxic language, while preserving common-sense zero-shot abilities, holds across all scales. AurA can be combined with pre-prompting strategies, boosting its average mitigation potential from $1.28\times$ to $2.35\times$. Moreover, AurA can counteract adversarial pre-prompts that maliciously elicit toxic content, making it an effective method for deploying safer and less toxic models.

Generating realistic lip motion from audio to simulate speech production is critical for driving natural character animation. Previous research has shown that traditional metrics used to optimize and assess models for generating lip motion from speech are not a good indicator of subjective opinion of animation quality. Devising metrics that align with subjective opinion first requires understanding what impacts human perception of quality. In this work, we focus on the degree of articulation and run a series of experiments to study how articulation strength impacts human perception of lip motion accompanying speech. Specifically, we study how increasing under-articulated (dampened) and over-articulated (exaggerated) lip motion affects human perception of quality. We examine the impact of articulation strength on human perception when considering only lip motion, where viewers are presented with talking faces represented by landmarks, and in the context of embodied characters, where viewers are presented with photo-realistic videos. Our results show that viewers prefer over-articulated lip motion consistently more than under-articulated lip motion and that this preference generalizes across different speakers and embodiments.

We present Spatial LibriSpeech, a spatial audio dataset with over 650 hours of 19-channel audio, first-order ambisonics, and optional distractor noise. Spatial LibriSpeech is designed for machine learning model training, and it includes labels for source position, speaking direction, room acoustics and geometry. Spatial LibriSpeech is generated by augmenting LibriSpeech samples with 200k+ simulated acoustic conditions across 8k+ synthetic rooms. To demonstrate the utility of our dataset, we train models on four spatial audio tasks, resulting in a median absolute error of 6.60{\deg} on 3D source localization, 0.43m on distance, 90.66ms on T30, and 2.74dB on DRR estimation. We show that the same models generalize well to widely-used evaluation datasets, e.g., obtaining a median absolute error of 12.43{\deg} on 3D source localization on TUT Sound Events 2018, and 157.32ms on T30 estimation on ACE Challenge.

In this paper we aim to investigate the mechanisms that guide text generation with pre-trained Transformer-based Language Models (TLMs). Grounded on the Product of Experts formulation by Hinton (1999), we describe a generative mechanism that exploits expert units which naturally exist in TLMs. Such units are responsible for detecting concepts in the input and conditioning text generation on such concepts. We describe how to identify expert units and how to activate them during inference in order to induce any desired concept in the generated output. We find that the activation of a surprisingly small amount of units is sufficient to steer text generation (as little as 3 units in a model with 345M parameters). While the objective of this work is to learn more about how TLMs work, we show that our method is effective for conditioning without fine-tuning or using extra parameters, even on fine-grained homograph concepts. Additionally, we show that our method can be used to correct gender bias present in the output of TLMs and achieves gender parity for all evaluated contexts. We compare our method with FUDGE and PPLM-BoW, and show that our approach is able to achieve gender parity at a lower perplexity and better Self-BLEU score. The proposed method is accessible to a wide audience thanks to its simplicity and minimal compute needs. The findings in this paper are a step forward in understanding the generative mechanisms of TLMs.

Automatic speech recognition (ASR) is widely used in consumer electronics. ASR greatly improves the utility and accessibility of technology, but usually the output is only word sequences without punctuation. This can result in ambiguity in inferring user-intent. We first present a transformer-based approach for punctuation prediction that achieves 8% improvement on the IWSLT 2012 TED Task, beating the previous state of the art [1]. We next describe our multimodal model that learns from both text and audio, which achieves 8% improvement over the text-only algorithm on an internal dataset for which we have both the audio and transcriptions. Finally, we present an approach to learning a model using contextual dropout that allows us to handle variable amounts of future context at test time.

We describe our novel deep learning approach for driving animated faces using both acoustic and visual information. In particular, speech-related facial movements are generated using audiovisual information, and non-verbal facial movements are generated using only visual information. To ensure that our model exploits both modalities during training, batches are generated that contain audio-only, video-only, and audiovisual input features. The probability of dropping a modality allows control over the degree to which the model exploits audio and visual information during training. Our trained model runs in real-time on resource limited hardware (e.g. a smart phone), it is user agnostic, and it is not dependent on a potentially error-prone transcription of the speech. We use subjective testing to demonstrate: 1) the improvement of audiovisual-driven animation over the equivalent video-only approach, and 2) the improvement in the animation of speech-related facial movements after introducing modality dropout. Without modality dropout, viewers prefer audiovisual-driven animation in 51% of the test sequences compared with only 18% for video-driven. After introducing dropout viewer preference for audiovisual-driven animation increases to 74%, but decreases to 8% for video-only.

In this paper we introduce Principal Filter Analysis (PFA), an easy to use and effective method for neural network compression. PFA exploits the correlation between filter responses within network layers to recommend a smaller network that maintain as much as possible the accuracy of the full model. We propose two algorithms: the first allows users to target compression to specific network property, such as number of trainable variable (footprint), and produces a compressed model that satisfies the requested property while preserving the maximum amount of spectral energy in the responses of each layer, while the second is a parameter-free heuristic that selects the compression used at each layer by trying to mimic an ideal set of uncorrelated responses. Since PFA compresses networks based on the correlation of their responses we show in our experiments that it gains the additional flexibility of adapting each architecture to a specific domain while compressing. PFA is evaluated against several architectures and datasets, and shows considerable compression rates without compromising accuracy, e.g., for VGG-16 on CIFAR-10, CIFAR-100 and ImageNet, PFA achieves a compression rate of 8×, 3×, and 1.4× with an accuracy gain of 0.4%, 1.4% points, and 2.4% respectively. Our tests show that PFA is competitive with state-of-the-art approaches while removing adoption barriers thanks to its practical implementation, intuitive philosophy and ease of use.

Speech-driven visual speech synthesis involves mapping acoustic speech features to the corresponding lip animation controls for a face model. This mapping can take many forms, but a powerful approach is to use deep neural networks (DNNs). The lack of synchronized audio, video, and depth data is a limitation to reliably train DNNs, especially for speaker-independent models. In this paper, we investigate adapting an automatic speech recognition (ASR) acoustic model (AM) for the visual speech synthesis problem. We train the ASR-AM on ten thousand hours of audio-only transcribed speech. The ASR-AM is then adapted to the visual speech synthesis domain using ninety hours of synchronized audio-visual speech. Using a subjective assessment test, we compared the performance of the AM-initialized DNN to a randomly initialized model. The results show that viewers significantly prefer animations generated from the AM-initialized DNN than the ones generated using the randomly initialized model. We conclude that visual speech synthesis can significantly benefit from the powerful representation of speech in the ASR acoustic models.

We describe experiments towards building a conversational digital assistant that considers the preferred conversational style of the user. In particular, these experiments are designed to measure whether users prefer and trust an assistant whose conversational style matches their own. To this end we conducted a user study where subjects interacted with a digital assistant that responded in a way that either matched their conversational style, or did not. Using self-reported personality attributes and subjects' feedback on the interactions, we built models that can reliably predict a user's preferred conversational style.

The problem of figure–ground segmentation is of great importance in both video editing and visual perception tasks. Classical video segmentation algorithms approach the problem from one of two perspectives. At one extreme, global approaches constrain the camera motion to simplify the image structure. At the other extreme, local approaches estimate motion in small image regions over a small number of frames and tend to produce noisy signals that are difficult to segment. With recent advances in image segmentation showing that sparse information is often sufficient for figure– ground segmentation it seems surprising then that with the extra temporal information of video, an unconstrained automatic figure–ground segmentation algorithm still eludes the research community. In this paper we present an automatic video segmentation algorithm that is intermediate between these two extremes and uses spatiotemporal features to regularize the segmentation. Detecting spatiotemporal T-junctions that indicate occlusion edges, we learn an occlusion edge model that is used within a colour contrast sensitive MRF to segment individual frames of a video sequence. T-junctions are learnt and classified using a support vector machine and a Gaussian mixture model is fitted to the (foreground, background) pixel pairs sampled from the detected T-junctions. Graph cut is then used to segment each frame of the video showing that sparse occlusion edge information can automatically initialize the video segmentation problem. 1

The goal of motion segmentation and layer extraction can be viewed as the detection and localization of occluding surfaces. A feature that has been shown to be a particularly strong indicator of occlusion, in both computer vision and neuroscience, is the T-junction; however, little progress has been made in T-junction detection. One reason for this is the difficulty in distinguishing false T-junctions (i.e. those not on an occluding edge) and real T-junctions in cluttered images. In addition to this, their photometric profile alone is not enough for reliable detection. This paper overcomes the first problem by searching for T-junctions not in space, but in space-time. This removes many false T-junctions and creates a simpler image structure to explore. The second problem is mitigated by learning the appearance of T-junctions in these spatiotemporal images. An RVM T-junction classifier is learnt from hand-labelled data using SIFT to capture its redundancy. This detector is then demonstrated in a novel occlusion detector that fuses Canny edges and T-junctions in the spatiotemporal domain to detect occluding edges in the spatial domain.

About 1.17 million people die in road crashes around the world each year. It is estimated that up to 30% of these fatalities are caused by fatigue and inattention. There are systems able to detect what is happening outside of the car, e.g., lane tracking, obstacle detection, pedestrian detection etc. Further on, there are also means for monitoring the actions of the driver. A natural step is to fuse the available data from within and outside of the car, and suggest a suitable response. This paper discusses driver assistance systems, lists a set of necessary core competencies of such a system and in particular presents a system for force-feedback in the steering wheel when crossing lanes. The presented system utilises a robust lane tracker which is experimentally evaluated for the purpose of driver assistance. In addition, preliminary results from simultaneous driver monitoring and lane tracking are presented that indicates a good correlation between the two, i.e. the driver's gaze direction and the structure of the road. These data can in turn be used for more advanced driver assistance systems in the future.