Neuromorphic computing shows promise for advancing computing efficiency and capabilities of AI applications using brain-inspired principles. However, the neuromorphic research field currently lacks standardized benchmarks, making it difficult to accurately measure technological advancements, compare performance with conventional methods, and identify promising future research directions. This article presents NeuroBench, a benchmark framework for neuromorphic algorithms and systems, which is collaboratively designed from an open community of researchers across industry and academia. NeuroBench introduces a common set of tools and systematic methodology for inclusive benchmark measurement, delivering an objective reference framework for quantifying neuromorphic approaches in both hardware-independent and hardware-dependent settings. For latest project updates, visit the project website (neurobench.ai).
Sequence learning, prediction and replay have been proposed to constitute the universal computations performed by the neocortex. The Hierarchical Temporal Memory (HTM) algorithm realizes these forms of computation. It learns sequences in an unsupervised and continuous manner using local learning rules, permits a context specific prediction of future sequence elements, and generates mismatch signals in case the predictions are not met. While the HTM algorithm accounts for a number of biological features such as topographic receptive fields, nonlinear dendritic processing, and sparse connectivity, it is based on abstract discrete-time neuron and synapse dynamics, as well as on plasticity mechanisms that can only partly be related to known biological mechanisms. Here, we devise a continuous-time implementation of the temporal-memory (TM) component of the HTM algorithm, which is based on a recurrent network of spiking neurons with biophysically interpretable variables and parameters. The model learns high-order sequences by means of a structural Hebbian synaptic plasticity mechanism supplemented with a rate-based homeostatic control. In combination with nonlinear dendritic input integration and local inhibitory feedback, this type of plasticity leads to the dynamic self-organization of narrow sequence-specific subnetworks. These subnetworks provide the substrate for a faithful propagation of sparse, synchronous activity, and, thereby, for a robust, context specific prediction of future sequence elements as well as for the autonomous replay of previously learned sequences. By strengthening the link to biology, our implementation facilitates the evaluation of the TM hypothesis based on experimentally accessible quantities. The continuous-time implementation of the TM algorithm permits, in particular, an investigation of the role of sequence timing for sequence learning, prediction and replay. We demonstrate this aspect by studying the effect of the sequence speed on the sequence learning performance and on the speed of autonomous sequence replay.
Sequence learning and prediction are considered principle computations performed by biological brains. Machine learning algorithms solve this type of task, but they require large amounts of training data and a substantial energy budget. An approach to overcome these issues and enable sequence learning with brain-like performance is neuromorphic hardware with brain-inspired learning algorithms. The Hierarchical Temporal Memory (HTM) is an algorithm inspired by the working principles of the neocortex and is able to learn and predict continuous sequences of elements. In a previous study, we showed that memristive devices, an emerging non-volatile memory technology, that is considered for energy efficient neuromorphic hardware, can be used as synapses in a biologically plausible version of the temporal memory algorithm of the HTM model. We subsequently presented a simulation study of an analog-mixed signal memristive hardware architecture that can implement the temporal learning algorithm. This architecture, which we refer to as MemSpikingTM, is based on a memristive crossbar array and a control circuitry implementing the neurons and the learning mechanism. In the study presented here, we demonstrate the functionality of the MemSpikingTM algorithm on a real memristive crossbar array, taped out in a commercially available 130nm CMOS technology node co-integrated with HfO based memristive devices. We explain the algorithm and the functionality of the crossbar array and peripheral circuitry and finally demonstrate context-dependent sequence learning using high-order sequences.
14 October 2022 in Frontiers in Integrative Neuroscience
Learning and replaying spatiotemporal sequences are fundamental computations performed by the brain and specifically the neocortex. These features are critical for a wide variety of cognitive functions, including sensory perception and the execution of motor and language skills. Although several computational models demonstrate this capability, many are either hard to reconcile with biological findings or have limited functionality. To address this gap, a recent study proposed a biologically plausible model based on a spiking recurrent neural network supplemented with read-out neurons. After learning, the recurrent network develops precise switching dynamics by successively activating and deactivating small groups of neurons. The read-out neurons are trained to respond to particular groups and can thereby reproduce the learned sequence. For the model to serve as the basis for further research, it is important to determine its replicability. In this Brief Report, we give a detailed description of the model and identify missing details, inconsistencies or errors in or between the original paper and its reference implementation. We re-implement the full model in the neural simulator NEST in conjunction with the NESTML modeling language and confirm the main findings of the original work.
Animals rely on different decision strategies when faced with ambiguous or uncertain cues. Depending on the context, decisions may be biased towards events that were most frequently experienced in the past, or be more explorative. A particular type of decision making central to cognition is sequential memory recall in response to ambiguous cues. A previously developed spiking neuronal network implementation of sequence prediction and recall learns complex, high-order sequences in an unsupervised manner by local, biologically inspired plasticity rules. In response to an ambiguous cue, the model deterministically recalls the sequence shown most frequently during training. Here, we present an extension of the model enabling a range of different decision strategies. In this model, explorative behavior is generated by supplying neurons with noise. As the model relies on population encoding, uncorrelated noise averages out, and the recall dynamics remain effectively deterministic. In the presence of locally correlated noise, the averaging effect is avoided without impairing the model performance, and without the need for large noise amplitudes. We investigate two forms of correlated noise occurring in nature: shared synaptic background inputs, and random locking of the stimulus to spatiotemporal oscillations in the network activity. Depending on the noise characteristics, the network adopts various recall strategies. This study thereby provides potential mechanisms explaining how the statistics of learned sequences affect decision making, and how decision strategies can be adjusted after learning.
2 May 2023 in Neuromorphic Computing and Engineering
Machine learning models for sequence learning and processing often suffer from high energy consumption and require large amounts of training data. The brain presents more efficient solutions to how these types of tasks can be solved. While this has inspired the conception of novel brain-inspired algorithms, their realizations remain constrained to conventional von-Neumann machines. Therefore, the potential power efficiency of the algorithm cannot be exploited due to the inherent memory bottleneck of the computing architecture. Therefore, we present in this paper a dedicated hardware implementation of a biologically plausible version of the Temporal Memory component of the Hierarchical Temporal Memory concept. Our implementation is built on a memristive crossbar array and is the result of a hardware-algorithm co-design process. Rather than using the memristive devices solely for data storage, our approach leverages their specific switching dynamics to propose a formulation of the peripheral circuitry, resulting in a more efficient design. By combining a brain-like algorithm with emerging non-volatile memristive device technology we strive for maximum energy efficiency. We present simulation results on the training of complex high-order sequences and discuss how the system is able to predict in a context-dependent manner. Finally, we investigate the energy consumption during the training and conclude with a discussion of scaling prospects.