We introduce two new source coding problems: robust sequential coding and robust predictive coding. For the Gauss-Markov source model with the mean squared error distortion measure, we characterize certain supporting ...
详细信息
We introduce two new source coding problems: robust sequential coding and robust predictive coding. For the Gauss-Markov source model with the mean squared error distortion measure, we characterize certain supporting hyperplanes of the rate region of these two coding problems. Our investigation also reveals an information-theoretic minimax theorem and the associated extremal inequalities.
Semantic representations in higher sensory cortices form the basis for robust, yet flexible behavior. These representations are acquired over the course of development in an unsupervised fashion and continuously maint...
详细信息
Semantic representations in higher sensory cortices form the basis for robust, yet flexible behavior. These representations are acquired over the course of development in an unsupervised fashion and continuously maintained over an organism's lifespan. predictive processing theories propose that these representations emerge from predicting or reconstructing sensory inputs. However, brains are known to generate virtual experiences, such as during imagination and dreaming, that go beyond previously experienced inputs. Here, we suggest that virtual experiences may be just as relevant as actual sensory inputs in shaping cortical representations. In particular, we discuss two complementary learning principles that organize representations through the generation of virtual experiences. First, "adversarial dreaming" proposes that creative dreams support a cortical implementation of adversarial learning in which feedback and feedforward pathways engage in a productive game of trying to fool each other. Second, "contrastive dreaming" proposes that the invariance of neuronal representations to irrelevant factors of variation is acquired by trying to map similar virtual experiences together via a contrastive learning process. These principles are compatible with known cortical structure and dynamics and the phenomenology of sleep thus providing promising directions to explain cortical learning beyond the classical predictive processing paradigm.
Our wellness relies on continuous interactions between our brain and body: different organs relay their current state to the brain and are regulated, in turn, by descending visceromotor commands from our brain and by ...
详细信息
Our wellness relies on continuous interactions between our brain and body: different organs relay their current state to the brain and are regulated, in turn, by descending visceromotor commands from our brain and by actions such as eating, drinking, thermotaxis, and predator escape. Human neuroimaging and theoretical studies suggest a key role for predictive processing by insular cortex in guiding these efforts to maintain bodily homeostasis. Here, we review recent studies recording and manipulating cellular activity in rodent insular cortex at timescales from seconds to hours. We argue that consideration of these findings in the context of predictive processing of future bodily states may reconcile several apparent discrepancies and offer a unifying, heuristic model for guiding future work.
Mismatch negativity (MMN) is an electrophysiological signature that occurs in response to unexpected stimuli. It is often referred to as a measure of memory-based change detection, because the elicitation of a predict...
详细信息
Mismatch negativity (MMN) is an electrophysiological signature that occurs in response to unexpected stimuli. It is often referred to as a measure of memory-based change detection, because the elicitation of a prediction error response relies on the formation of a prediction, which in turn, is dependent upon intact memory of previous auditory stimulation. As such, the MMN is altered in conditions in which memory is affected, such as Alzheimer's disease, schizophrenia and healthy aging. The most prominent pharmacological finding for MMN strengthens the link between MMN and synaptic plasticity, as glutamate N-methyl-D-aspartate receptor (NMDA-R) antagonists reduce the MMN response. However, recent data has begun to demonstrate that the link between NMDA-R function and MMN is not as clear as once thought, with low dose and low affinity NMDA-R antagonists observed to facilitate MMN. (c) 2020 IBRO. Published by Elsevier Ltd. All rights reserved.
Inspired by Ronald Giere's (1989, 1992) cognitive approach to scientific models, Cognitive Structural Realism (CSR) has presented a naturalist account of scientific representation (Beni, 2019a). CSR characterises ...
详细信息
Inspired by Ronald Giere's (1989, 1992) cognitive approach to scientific models, Cognitive Structural Realism (CSR) has presented a naturalist account of scientific representation (Beni, 2019a). CSR characterises the structure of theories in terms of cognitive structures. These are informational structures embodied in the brains of (allegedly individual) scientists. CSR accounts for scientific representation in terms of the dynamical relationship between the organism and its environment. The proposal has been criticised on account of its negligence of social aspects of scientific practice. The present paper aims to chart out a reply to the objection. It shows that cognitive structures do not need to be put inside the brains of single individuals. Cognitive structures are redefined as extended structures in distributed cognitive systems (such as a scientific group) under Free Energy Principle.
Self-supervised learning (SSL) is an approach to pretrain deep networks with unlabeled datasets by using pretext tasks that use images as “ground truth”. Pretext tasks have been shown to impact accuracy of task cate...
详细信息
ISBN:
(数字)9798331520526
ISBN:
(纸本)9798331520533
Self-supervised learning (SSL) is an approach to pretrain deep networks with unlabeled datasets by using pretext tasks that use images as “ground truth”. Pretext tasks have been shown to impact accuracy of task categories, e.g. segmentation vs. classification. However, versatility of SSL features to downstream tasks involving different modalities has not been studied. We benchmarked impact of SSL tasks such as contrastive predictive coding, token self-distillation, and generative masked image modeling (MIM) with 3D vision transformer performed using 10K 3D-CTs (or 1.89M images) from various disease sites. SSL pretraining was used to assess (a) multi-organ segmentation under data-limited fine tuning, (b) feature reuse and (c) organ localization with multi-head attention. Analysis showed that pretext tasks combining MIM and token self-distillation balanced local and global attention distance, produced higher segmentation accuracy in few-shot and data-limited settings for MRI and CT. Feature reuse was impacted by similarity of pretraining and fine-tuning modality.
Due to the abundance of the new digital media data, the issue of image quality and volume of data requiring compression has become a significant factor of concern, especially in media storage and transmitting. This wo...
详细信息
ISBN:
(数字)9798331508456
ISBN:
(纸本)9798331508463
Due to the abundance of the new digital media data, the issue of image quality and volume of data requiring compression has become a significant factor of concern, especially in media storage and transmitting. This work affords a comparative analysis of different image compression techniques with focus on the compression ratio, quality preservation, and complexity. A new hybrid model of predictive coding, Run Length coding and Quantum Entropy coding (QEC) is proposed and shown to exhibit negligible quality loss with substantial space saving. The experimental outcomes show that the proposed method reduces space 80 percent and works better than previous methods for areas requiring high speed and relative accuracy. These insights are timely, as practical computing-communication trade-offs are paramount in the new generation of social networks, medicine, and multimedia streaming.
Deep predictive coding networks (DPCNs) effectively model and capture video features through a bi-directional inference without labels. They are based on an overcomplete description of video scenes, and one of the bot...
详细信息
ISBN:
(数字)9798350368741
ISBN:
(纸本)9798350368758
Deep predictive coding networks (DPCNs) effectively model and capture video features through a bi-directional inference without labels. They are based on an overcomplete description of video scenes, and one of the bottlenecks has been the lack of effective sparsification techniques to find discriminative and robust dictionaries. This paper proposes a DPCN with a fast inference of internal dictionaries and variables that achieve high sparsity and feature clustering accuracy. The proposed unsupervised learning procedure uses majorization-minimization (MM) to smooth sparsity constraints in optimization and admits explainability and convergence. Experiments in the image and video data sets CIFAR-10, Super Mario Bros, and Coil-100 validate that the approach outperforms previous versions of DPCNs on learning rate, sparsity ratio, and feature clustering accuracy. This advance opens the door for general applications in object recognition in video without labels.
Research into the neural foundation of perception asserts a model where top-down predictions modulate the bottom-up processing of sensory input. Despite becoming increasingly influential in cognitive neuroscience, the...
详细信息
Research into the neural foundation of perception asserts a model where top-down predictions modulate the bottom-up processing of sensory input. Despite becoming increasingly influential in cognitive neuroscience, the precise account of this predictive coding framework remains debated. In this study, we aim to contribute to this debate by investigating how predictions about prosody facilitate speech perception, and to shed light especially on lexical access influenced by simultaneous predictions in different domains, inter alia, prosodic and semantic. Using a passive auditory oddball paradigm, we examined neural responses to prosodic changes, leading to a semantic change as in Dutch nouns canon [' kaMODIFIER LETTER TRIANGULAR COLONnon] 'canon' vs kanon [kaMODIFIER LETTER TRIANGULAR COLON ' non] 'cannon', and used acoustically identical pseudowords as controls. Results from twenty-eight native speakers of Dutch (age range 18-32 years) indicated an enhanced P50/N100 complex to prosodic change in pseudowords as well as an MMN response to both words and pseudowords. The enhanced P50/N100 response to pseudowords is claimed to indicate that all relevant auditory information is still processed by the brain, whereas the reduced response to words might reflect the suppression of information that has already been encoded. The MMN response to pseudowords and words, on the other hand, is best justified by the unification of previously established prosodic representations with sensory and semantic input respectively. This pattern of results is in line with the predictive coding framework acting on multiple levels and is of crucial importance to indicate that predictions about linguistic prosodic information are utilized by the brain as early as 50 ms.
Electroencephalography (EEG) based emotion recognition shows promise in human-computer interaction and mental health monitoring, but faces challenges in cross-dataset generalization. This study introduces the Unified ...
详细信息
ISBN:
(数字)9798331540067
ISBN:
(纸本)9798331540074
Electroencephalography (EEG) based emotion recognition shows promise in human-computer interaction and mental health monitoring, but faces challenges in cross-dataset generalization. This study introduces the Unified Framework for EEG-based Emotion Recognition with Self-Supervised Learning (UNiFY-ESSL), combining self-supervised learning (SSL) and advanced sampling techniques to integrate multiple EEG datasets. UNiFY-ESSL utilizes a unified 14-channel setup across all datasets (SEED, DEAP, and DREAMER), allowing consistent preprocessing and feature extraction. This approach adapts datasets with varying initial channel counts (62 for SEED, 32 for DEAP, and 14 for DREAMER) to a common configuration based on DREAMER's channels. The framework explores contrastive learning methods, specifically Simple Contrastive Learning (Sim-CLR) and Contrastive predictive coding (CPC), enhanced by a novel stratified sampling scheme. Experiments yield impressive results: SimCLR achieves F1-scores of 82.62%, 87.83%, and 89.05% for SEED, DEAP, and DREAMER respectively, while CPC attains 81.35%, 82.27%, and 91.23%. It demonstrates improved cross-dataset generalization, with a 1–2% performance gain on DREAMER and maintained performance on DEAP despite channel reduction. However, SEED experiences a 3% F1-score drop due to significant channel reduction. These findings highlight the potential of our sampling method in advancing EEG-based emotion recognition while underscoring the need for future work on efficient channel combination strategies, particularly for datasets with higher initial channel counts.
暂无评论