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 learning 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 learning paradigm.

Although artificial intelligence (AI) systems have been shown to improve the accuracy of initial melanoma diagnosis, the lack of transparency in how these systems identify melanoma poses severe obstacles to user acceptance. Explainable artificial intelligence (XAI) methods can help to increase transparency, but most XAI methods are unable to produce precisely located domain-specific explanations, making the explanations difficult to interpret. Moreover, the impact of XAI methods on dermatologists has not yet been evaluated. Extending on two existing classifiers, we developed an XAI system that produces text and region based explanations that are easily interpretable by dermatologists alongside its differential diagnoses of melanomas and nevi. To evaluate this system, we conducted a three-part reader study to assess its impact on clinicians' diagnostic accuracy, confidence, and trust in the XAI-support. We showed that our XAI's explanations were highly aligned with clinicians' explanations and that both the clinicians' trust in the support system and their confidence in their diagnoses were significantly increased when using our XAI compared to using a conventional AI system. The clinicians' diagnostic accuracy was numerically, albeit not significantly, increased. This work demonstrates that clinicians are willing to adopt such an XAI system, motivating their future use in the clinic.

Humans and other animals learn to extract general concepts from sensory experience without extensive teaching. This ability is thought to be facilitated by offline states like sleep where previous experiences are systemically replayed. However, the characteristic creative nature of dreams suggests that learning semantic representations may go beyond merely replaying previous experiences. We support this hypothesis by implementing a cortical architecture inspired by generative adversarial networks (GANs). Learning in our model is organized across three different global brain states mimicking wakefulness, NREM and REM sleep, optimizing different, but complementary objective functions. We train the model on standard datasets of natural images and evaluate the quality of the learned representations. Our results suggest that generating new, virtual sensory inputs via adversarial dreaming during REM sleep is essential for extracting semantic concepts, while replaying episodic memories via perturbed dreaming during NREM sleep improves the robustness of latent representations. The model provides a new computational perspective on sleep states, memory replay and dreams and suggests a cortical implementation of GANs.

Large language models (LLMs) are now widely accessible, reaching learners at all educational levels. This development has raised concerns that their use may circumvent essential learning processes and compromise the integrity of established assessment formats. In physics education, where problem solving plays a central role in instruction and assessment, it is therefore essential to understand the physics-specific problem-solving capabilities of LLMs. Such understanding is key to informing responsible and pedagogically sound approaches to integrating LLMs into instruction and assessment. This study therefore compares the problem-solving performance of a general-purpose LLM (GPT-4o, using varying prompting techniques) and a reasoning-optimized model (o1-preview) with that of participants of the German Physics Olympiad, based on a set of well-defined Olympiad problems. In addition to evaluating the correctness of the generated solutions, the study analyzes characteristic strengths and limitations of LLM-generated solutions. The findings of this study indicate that both tested LLMs (GPT-4o and o1-preview) demonstrate advanced problem-solving capabilities on Olympiad-type physics problems, on average outperforming the human participants. Prompting techniques had little effect on GPT-4o's performance, while o1-preview almost consistently outperformed both GPT-4o and the human benchmark. Based on these findings, the study discusses implications for the design of summative and formative assessment in physics education, including how to uphold assessment integrity and support students in critically engaging with LLMs.

We present results from a new approach to learning and plasticity in neuromorphic hardware systems: to enable flexibility in implementable learning mechanisms while keeping high efficiency associated with neuromorphic implementations, we combine a general-purpose processor with full-custom analog elements. This processor is operating in parallel with a fully parallel neuromorphic system consisting of an array of synapses connected to analog, continuous time neuron circuits. Novel analog correlation sensor circuits process spike events for each synapse in parallel and in real-time. The processor uses this pre-processing to compute new weights possibly using additional information following its program. Therefore, learning rules can be defined in software giving a large degree of flexibility. Synapses realize correlation detection geared towards Spike-Timing Dependent Plasticity (STDP) as central computational primitive in the analog domain. Operating at a speed-up factor of 1000 compared to biological time-scale, we measure time-constants from tens to hundreds of micro-seconds. We analyze variability across multiple chips and demonstrate learning using a multiplicative STDP rule. We conclude, that the presented approach will enable flexible and efficient learning as a platform for neuroscientific research and technological applications.

The neuromorphic BrainScaleS-2 ASIC comprises mixed-signal neurons and synapse circuits as well as two versatile digital microprocessors. Primarily designed to emulate spiking neural networks, the system can also operate in a vector-matrix multiplication and accumulation mode for artificial neural networks. Analog multiplication is carried out in the synapse circuits, while the results are accumulated on the neurons' membrane capacitors. Designed as an analog, in-memory computing device, it promises high energy efficiency. Fixed-pattern noise and trial-to-trial variations, however, require the implemented networks to cope with a certain level of perturbations. Further limitations are imposed by the digital resolution of the input values (5 bit), matrix weights (6 bit) and resulting neuron activations (8 bit). In this paper, we discuss BrainScaleS-2 as an analog inference accelerator and present calibration as well as optimization strategies, highlighting the advantages of training with hardware in the loop. Among other benchmarks, we classify the MNIST handwritten digits dataset using a two-dimensional convolution and two dense layers. We reach 98.0% test accuracy, closely matching the performance of the same network evaluated in software.

Based on a sample of (10.09$\pm$0.04)$\times$10$^{9}$ $J/\psi$ events collected with the BESIII detector operating at the BEPCII storage ring, a partial wave analysis of the decay $J/\psi \rightarrow \gamma\eta\eta'$ is performed. An isoscalar state with exotic quantum numbers $J^{PC}=1^{-+}$, denoted as $\eta_1(1855)$, has been observed for the first time with statistical significance larger than 19$\sigma$. Its mass and width are measured to be (1855$\pm$9$_{-1}^{+6}$)~MeV/$c^{2}$ and (188$\pm$18$_{-8}^{+3}$)~MeV, respectively. The product branching fraction ${\cal B}(J/\psi$$\rightarrow$$ \gamma\eta_1(1855)$$\rightarrow$$\gamma\eta\eta')$is measured to be (2.70$\pm 0.41 _{-0.35}^{+0.16}) \times$10$^{-6}$. The first uncertainties are statistical and the second are systematic. In addition, an upper limit on the branching ratio ${\cal B}(f_0(1710)$$\rightarrow$$\eta\eta')$/${\cal B}(f_0(1710)$$\rightarrow$$\pi\pi)$is determined to be$1.61 \times 10^{-3}$ at 90\% confidence level, which lends support to the hypothesis that the $f_0(1710)$ has a large glueball component.

Traditional neuromorphic hardware architectures rely on event-driven computation, where the asynchronous transmission of events, such as spikes, triggers local computations within synapses and neurons. While machine learning frameworks are commonly used for gradient-based training, their emphasis on dense data structures poses challenges for processing asynchronous data such as spike trains. This problem is particularly pronounced for typical tensor data structures. In this context, we present a novel library (jaxsnn) built on top of JAX, that departs from conventional machine learning frameworks by providing flexibility in the data structures used and the handling of time, while maintaining Autograd functionality and composability. Our library facilitates the simulation of spiking neural networks and gradient estimation, with a focus on compatibility with time-continuous neuromorphic backends, such as the BrainScaleS-2 system, during the forward pass. This approach opens avenues for more efficient and flexible training of spiking neural networks, bridging the gap between traditional neuromorphic architectures and contemporary machine learning frameworks.

The AMoRE (Advanced Mo-based Rare process Experiment) project is a series of experiments that use advanced cryogenic techniques to search for the neutrinoless double-beta decay of \mohundred. The work is being carried out by an international collaboration of researchers from eight countries. These searches involve high precision measurements of radiation-induced temperature changes and scintillation light produced in ultra-pure \Mo[100]-enriched and \Ca[48]-depleted calcium molybdate ($\mathrm{^{48depl}Ca^{100}MoO_4}$) crystals that are located in a deep underground laboratory in Korea. The \mohundred nuclide was chosen for this \zeronubb decay search because of its high $Q$-value and favorable nuclear matrix element. Tests have demonstrated that \camo crystals produce the brightest scintillation light among all of the molybdate crystals, both at room and at cryogenic temperatures. $\mathrm{^{48depl}Ca^{100}MoO_4}$ crystals are being operated at milli-Kelvin temperatures and read out via specially developed metallic-magnetic-calorimeter (MMC) temperature sensors that have excellent energy resolution and relatively fast response times. The excellent energy resolution provides good discrimination of signal from backgrounds, and the fast response time is important for minimizing the irreducible background caused by random coincidence of two-neutrino double-beta decay events of \mohundred nuclei. Comparisons of the scintillating-light and phonon yields and pulse shape discrimination of the phonon signals will be used to provide redundant rejection of alpha-ray-induced backgrounds. An effective Majorana neutrino mass sensitivity that reaches the expected range of the inverted neutrino mass hierarchy, i.e., 20-50 meV, could be achieved with a 200~kg array of $\mathrm{^{48depl}Ca^{100}MoO_4}$ crystals operating for three years.

We present software facilitating the usage of the BrainScaleS-2 analog neuromorphic hardware system as an inference accelerator for artificial neural networks. The accelerator hardware is transparently integrated into the PyTorch machine learning framework using its extension interface. In particular, we provide accelerator support for vector-matrix multiplications and convolutions; corresponding software-based autograd functionality is provided for hardware-in-the-loop training. Automatic partitioning of neural networks onto one or multiple accelerator chips is supported. We analyze implementation runtime overhead during training as well as inference, provide measurements for existing setups and evaluate the results in terms of the accelerator hardware design limitations. As an application of the introduced framework, we present a model that classifies activities of daily living with smartphone sensor data.

In this article, a chiral plasmonic hydrogen-sensing platform using palladium-based nanohelices is demonstrated. Such 3D chiral nanostructures fabricated by nanoglancing angle deposition exhibit strong circular dichroism both experimentally and theoretically. The chiroptical properties of the palladium nanohelices are altered upon hydrogen uptake and sensitively depend on the hydrogen concentration. Such properties are well suited for remote and spark-free hydrogen sensing in the flammable range. Hysteresis is reduced, when an increasing amount of gold is utilized in the palladium-gold hybrid helices. As a result, the linearity of the circular dichroism in response to hydrogen is significantly improved. The chiral plasmonic sensor scheme is of potential interest for hydrogen-sensing applications, where good linearity and high sensitivity are required.

We propose the BayesDose-Framework, a Bayesian approach for fast and accurate dose prediction in proton therapy. Our framework is based on a previously published deterministic LSTM model and is trained and evaluated on simulated beamlet doses from water phantoms and patient geometries. We parameterize the network's weights using 2D Gaussian mixture models and use ensemble predictions to quantify mean dose predictions and their standard deviation. The BayesDose model performs similarly to the deterministic variant. The uncertainty predictions are conservative but correlate well spatially and in magnitude with dose differences. This correlation is reduced when applied to patient data with unseen relative stopping power value ranges, which could be successfully addressed by re-training. We parallelize predictions and presample network weights to reduce runtime overhead. Bayesian models like BayesDose can provide fast predictions with quality equal to deterministic models and may support decision making and quality assurance in clinical settings in the future.